guides/parking-management/

[giscus[bot]]

guides/parking-management/

Parking Management System Using Ultralytics YOLOv8

https://docs.ultralytics.com/guides/parking-management/

[ilyasKerrouche]

Hi , when i run the first script i get this error:

File "C:\Users\explo\miniconda3\Lib\site-packages\ultralytics\solutions\parking_management.py", line 71, in upload_image
self.canvas.config(width=canvas_width, height=canvas_height)
^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'config'

How can i solve?

[ilyasKerrouche]

Hello! I suggest to fix the uplad_image function like this:
`

      def upload_image(self):
              """Upload an image and resize it to fit canvas."""
      
              
              self.image_path = filedialog.askopenfilename(filetypes=[("Image Files", "*.png;*.jpg;*.jpeg")])
              if not self.image_path:
                  return
      
              self.image = Image.open(self.image_path)
              self.img_width, self.img_height = self.image.size
      
              # Calculate the aspect ratio and resize image
              aspect_ratio = self.img_width / self.img_height
              if aspect_ratio > 1:
                  # Landscape orientation
                  canvas_width = min(self.canvas_max_width, self.img_width)
                  canvas_height = int(canvas_width / aspect_ratio)
              else:
                  # Portrait orientation
                  canvas_height = min(self.canvas_max_height, self.img_height)
                  canvas_width = int(canvas_height * aspect_ratio)
      
             # Setup canvas for image display 
              self.canvas = tk.Canvas(self.master, bg="white", width=canvas_width, height=canvas_height)
              self.canvas.pack(side=tk.BOTTOM)
              self.canvas.bind("<Button-1>", self.on_canvas_click)
      
      
              self.canvas.config(width=canvas_width, height=canvas_height)
              resized_image = self.image.resize((canvas_width, canvas_height), Image.LANCZOS)
              self.canvas_image = ImageTk.PhotoImage(resized_image)
              self.canvas.create_image(0, 0, anchor=tk.NW, image=self.canvas_image)
      
              # Reset bounding boxes and current box
              self.bounding_boxes = []
              self.current_box = []`

And i tested the code with this video

But the model is not recognizing the cars (cup, microwave ecc instead)

Do you have some videos template to test the code?

[pderrenger]

@eXplosive57 hello! Thanks for sharing your updated upload_image function and the scenario you're experimenting with. I appreciate your attention to detail. 😊

Regarding the issue of the model not recognizing cars correctly and identifying other objects like cups or microwaves instead, it sounds like it might be a case of model misclassification, which can be due to a variety of factors:

1. Model Training: The model may not be trained sufficiently on images similar to those in your video. Consider retraining the model with a more representative dataset of parking lot images.
2. Quality and Size of the Input Images: Make sure the images are clear and of sufficient resolution. YOLO models generally perform better with higher-resolution inputs.
3. Model Configuration: Check if the right configuration and weights are being used. Ensure that the model is set up to detect objects relevant to parking management.

For testing purposes, you can use standard traffic or parking lot videos which are often used in object detection benchmarks, or you can capture your own dataset from typical parking areas. Here is an example of initializing and using the parking management model:

import cv2
from ultralytics.solutions.parking_management import ParkingManagement

# Ensure you have the correct model path and polygon JSON
management = ParkingManagement(model_path="yolov8n.pt")
polygon_json_path = "path/to/your/bounding_boxes.json"

cap = cv2.VideoCapture("path/to/your/video.mp4")
while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False)

    if results[0].and results[0].boxes:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)

cap.release()
cv2.destroyAllWindows()

This setup should help you streamline the detection process. If the issue persists, try fine-tuning the detection model or adjusting the confidence and IOU thresholds. Let me know how it goes or if you need further assistance!

[RizwanMunawar]

Thank you, @eXplosive57, for providing the prompt solution! I've been able to replicate the issue and will address it promptly. Regarding the model, training it on the visdrone dataset could enhance its ability to detect objects in drone or aerial views. Thanks

VisDrone Dataset Docs: https://docs.ultralytics.com/datasets/detect/visdrone/

[ilyasKerrouche]

Thanks for your answers! @RizwanMunawar @pderrenger

[RizwanMunawar]

@eXplosive57 This issue has been resolved! You can utilize the parking management system by installing the latest ultralytics PYPI package using the command mentioned below! pip install ultralytics

Thanks!

[Gxiaodiao]

Hello, I use the iOS system, when using the following code to meet the data standards, the image can not be loaded, pngd,jpg,jpeg can not be loaded, help me to answer, thank you

from ultralytics import solutions

solutions.ParkingPtsSelection()

[glenn-jocher]

@Gxiaodiao hello,

Thank you for reaching out with your issue regarding image loading in the iOS system using the solutions.ParkingPtsSelection() function.

To assist you better, could you please provide the following details:

1. The version of the Ultralytics library you are using.
2. Any error messages or logs that appear when you attempt to load the images.
3. The specific iOS version and device you are using.

This information will help us understand the problem more clearly and provide a more accurate solution. Meanwhile, ensure that the image paths are correct and accessible from your application.

Looking forward to your response!

[adiyazid]

Can it support camera from the side or angle instead from top view?

[pderrenger]

Yes, Ultralytics YOLOv8 can support camera angles from the side or at an angle, not just from a top view. The model's robust detection capabilities allow it to effectively identify vehicles in various orientations and angles, which can be particularly useful in dynamic parking environments. Adjusting the detection thresholds and fine-tuning the model with angle-specific training data can further enhance accuracy for side or angled views. If you have any more questions or need further assistance, feel free to ask! 🚗👍

[adiyazid]

Hi Paula, Thanks for your reply, at this moment I'm using a licensed subscription platform to detect the vehicle presence at the parking bays. I'd expect YOLOv8 could offer an open source solution which could reduce the license cost. Please provide an Example Workflow using YOLOv8 based on the use of a solar powered camera using ftp function to send parking bay images to a ftp server. Do reply to me soon, thank you. Adi Y Rosl

…

On Fri, Jun 7, 2024 at 9:13 PM Paula Derrenger ***@***.***> wrote: Yes, Ultralytics YOLOv8 can support camera angles from the side or at an angle, not just from a top view. The model's robust detection capabilities allow it to effectively identify vehicles in various orientations and angles, which can be particularly useful in dynamic parking environments. Adjusting the detection thresholds and fine-tuning the model with angle-specific training data can further enhance accuracy for side or angled views. If you have any more questions or need further assistance, feel free to ask! 🚗👍 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQXHTL6NTQV7HOZOAWLZGGWXFAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMBSGIZTO> . You are receiving this because you commented.Message ID: ***@***.*** com>

[glenn-jocher]

Hi Adi,

Thank you for reaching out! It's great to hear that you're considering YOLOv8 for your parking management needs. YOLOv8 can indeed provide a cost-effective, open-source solution for vehicle detection in parking bays. Below, I'll walk you through an example workflow to set up a parking management system using a solar-powered camera that sends images to an FTP server.

Example Workflow for Parking Management with YOLOv8

1. Setting Up the FTP Server

Ensure your solar-powered camera is configured to send images to your FTP server. You'll need the FTP server details (hostname, username, password, and the directory where images are stored).

2. Downloading Images from FTP Server

Use Python's ftplib library to download images from the FTP server.

import ftplib
import os

ftp = ftplib.FTP('ftp.yourserver.com')
ftp.login('username', 'password')
ftp.cwd('/path/to/images')

filenames = ftp.nlst()  # List of filenames in the directory

for filename in filenames:
    local_filename = os.path.join('/local/path/to/save/images', filename)
    with open(local_filename, 'wb') as file:
        ftp.retrbinary('RETR ' + filename, file.write)

ftp.quit()

3. Selecting Parking Points

Use the ParkingPtsSelection tool to define parking areas.

from ultralytics import solutions

solutions.ParkingPtsSelection()

This will allow you to create polygons around parking spaces and save the configuration as a JSON file.

4. Parking Management Code

Here's a complete example of how to set up the parking management system using YOLOv8:

import cv2
from ultralytics import solutions

# Path to the JSON file created with the point selection tool
polygon_json_path = "bounding_boxes.json"

# Initialize video capture from the local directory where images are saved
cap = cv2.VideoCapture("/local/path/to/save/images/video.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

# Video writer to save the output
video_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

# Initialize parking management object
management = solutions.ParkingManagement(model_path="yolov8n.pt")

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

Optional Arguments for Customization

You can customize the appearance and behavior of the parking management system using optional arguments:

management = solutions.ParkingManagement(
    model_path="yolov8n.pt",
    txt_color=(0, 0, 0),
    bg_color=(255, 255, 255),
    occupied_region_color=(0, 255, 0),
    available_region_color=(0, 0, 255),
    margin=10
)

This setup should help you get started with a cost-effective and efficient parking management system using YOLOv8. If you have any further questions or need additional assistance, feel free to ask! 🚗👍

[adiyazid]

Hi, I've deployed the Yolov8 to test it and I've got the error message when verifying the installation as follows. yolo task=detect mode=predict model=yolov8n.pt source=' https://ultralytics.com/images/zidane.jpg' Traceback (most recent call last): File "/yolov8env/bin/yolo", line 5, in <module> from ultralytics.cfg import entrypoint File "/yolov8env/lib/python3.12/site-packages/ultralytics/__init__.py", line 10, in <module> from ultralytics.data.explorer.explorer import Explorer File "/yolov8env/lib/python3.12/site-packages/ultralytics/data/__init__.py", line 3, in <module> from .base import BaseDataset File "/yolov8env/lib/python3.12/site-packages/ultralytics/data/base.py", line 12, in <module> import cv2 File "/yolov8env/lib/python3.12/site-packages/cv2/__init__.py", line 181, in <module> bootstrap() File "/yolov8env/lib/python3.12/site-packages/cv2/__init__.py", line 153, in bootstrap native_module = importlib.import_module("cv2") ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.12/importlib/__init__.py", line 90, in import_module return _bootstrap._gcd_import(name[level:], package, level) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ImportError: libGL.so.1: cannot open shared object file: No such file or directory I did try to search and install libGL.so.1 but cannot find it, Do reply to me soon, thanks.

…

On Sat, Jun 8, 2024 at 4:26 AM Glenn Jocher ***@***.***> wrote: Hi Adi, Thank you for reaching out! It's great to hear that you're considering YOLOv8 for your parking management needs. YOLOv8 can indeed provide a cost-effective, open-source solution for vehicle detection in parking bays. Below, I'll walk you through an example workflow to set up a parking management system using a solar-powered camera that sends images to an FTP server. Example Workflow for Parking Management with YOLOv8 1. *Setting Up the FTP Server* Ensure your solar-powered camera is configured to send images to your FTP server. You'll need the FTP server details (hostname, username, password, and the directory where images are stored). 2. *Downloading Images from FTP Server* Use Python's ftplib library to download images from the FTP server. import ftplibimport os ftp = ftplib.FTP('ftp.yourserver.com')ftp.login('username', 'password')ftp.cwd('/path/to/images') filenames = ftp.nlst() # List of filenames in the directory for filename in filenames: local_filename = os.path.join('/local/path/to/save/images', filename) with open(local_filename, 'wb') as file: ftp.retrbinary('RETR ' + filename, file.write) ftp.quit() 3. *Selecting Parking Points* Use the ParkingPtsSelection tool to define parking areas. from ultralytics import solutions solutions.ParkingPtsSelection() This will allow you to create polygons around parking spaces and save the configuration as a JSON file. 4. *Parking Management Code* Here's a complete example of how to set up the parking management system using YOLOv8: import cv2from ultralytics import solutions # Path to the JSON file created with the point selection toolpolygon_json_path = "bounding_boxes.json" # Initialize video capture from the local directory where images are savedcap = cv2.VideoCapture("/local/path/to/save/images/video.mp4")assert cap.isOpened(), "Error reading video file"w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) # Video writer to save the outputvideo_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Initialize parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") while cap.isOpened(): ret, im0 = cap.read() if not ret: break json_data = management.parking_regions_extraction(polygon_json_path) results = management.model.track(im0, persist=True, show=False) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) management.display_frames(im0) video_writer.write(im0) cap.release()video_writer.release()cv2.destroyAllWindows() Optional Arguments for Customization You can customize the appearance and behavior of the parking management system using optional arguments: management = solutions.ParkingManagement( model_path="yolov8n.pt", txt_color=(0, 0, 0), bg_color=(255, 255, 255), occupied_region_color=(0, 255, 0), available_region_color=(0, 0, 255), margin=10 ) This setup should help you get started with a cost-effective and efficient parking management system using YOLOv8. If you have any further questions or need additional assistance, feel free to ask! 🚗👍 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQWTFLZQL5J2E6MIANDZGIJQXAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMBXGE2DO> . You are receiving this because you commented.Message ID: ***@***.*** com>

[glenn-jocher]

Hi there,

Thank you for reaching out! It looks like you're encountering an issue with the libGL.so.1 library, which is a common dependency for OpenCV. This error typically occurs when the required OpenGL libraries are not installed on your system.

To resolve this issue, you can install the missing library using the following command, depending on your operating system:

For Ubuntu/Debian:

sudo apt-get update
sudo apt-get install -y libgl1-mesa-glx

For CentOS/RHEL:

sudo yum install -y mesa-libGL

For macOS, you can use Homebrew:

brew install libgl

After installing the necessary library, try running your YOLOv8 command again:

yolo task=detect mode=predict model=yolov8n.pt source='https://ultralytics.com/images/zidane.jpg'

This should resolve the libGL.so.1 error and allow you to proceed with your YOLOv8 deployment. If you encounter any further issues or have additional questions, feel free to ask! 😊

[adiyazid]

I got this message. Reading package lists... Done Building dependency tree... Done Reading state information... Done Package libgl1-mesa-glx is not available, but is referred to by another package. This may mean that the package is missing, has been obsoleted, or is only available from another source E: Package 'libgl1-mesa-glx' has no installation candidate I'm using Ubuntu 24.04 LTS

…

On Sat, Jun 8, 2024 at 7:43 PM Glenn Jocher ***@***.***> wrote: Hi there, Thank you for reaching out! It looks like you're encountering an issue with the libGL.so.1 library, which is a common dependency for OpenCV. This error typically occurs when the required OpenGL libraries are not installed on your system. To resolve this issue, you can install the missing library using the following command, depending on your operating system: For Ubuntu/Debian: sudo apt-get update sudo apt-get install -y libgl1-mesa-glx For CentOS/RHEL: sudo yum install -y mesa-libGL For macOS, you can use Homebrew: brew install libgl After installing the necessary library, try running your YOLOv8 command again: yolo task=detect mode=predict model=yolov8n.pt source='https://ultralytics.com/images/zidane.jpg' This should resolve the libGL.so.1 error and allow you to proceed with your YOLOv8 deployment. If you encounter any further issues or have additional questions, feel free to ask! 😊 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQRJAVVLN6DWS47LWW3ZGLU7NAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMJSHE2TS> . You are receiving this because you commented.Message ID: ***@***.*** com>

[pderrenger]

Hi there,

Thank you for your detailed report! It seems like you're encountering an issue with the libgl1-mesa-glx package on Ubuntu 24.04 LTS. This package is essential for OpenCV to function correctly, and its absence can cause the errors you're seeing.

To resolve this, you can try installing the libgl1 package, which should provide the necessary OpenGL libraries. Please run the following commands in your terminal:

sudo apt-get update
sudo apt-get install -y libgl1

If the issue persists, it might be due to changes or updates in the package repositories for Ubuntu 24.04 LTS. In such cases, you can also try installing the libgl1-mesa-glx package from an older Ubuntu release or using a different package source.

Additionally, please ensure that you are using the latest versions of torch and ultralytics. You can update them using:

pip install --upgrade torch ultralytics

Once you've installed the necessary libraries and updated your packages, try running your YOLOv8 command again:

yolo task=detect mode=predict model=yolov8n.pt source='https://ultralytics.com/images/zidane.jpg'

If you continue to experience issues, please provide a minimum reproducible code example so we can investigate further. You can find more information on creating a reproducible example here: Minimum Reproducible Example.

Feel free to reach out if you have any more questions or need further assistance! 😊

[adiyazid]

Hi, I've resolved the issue and tested the command. Please provide a detailed guide on how to use the ParkingPtsSelection tool to define parking areas. This will allow me to create polygons around parking spaces and save the configuration as a JSON file. Thanks.

…

On Sun, Jun 9, 2024 at 3:18 AM Paula Derrenger ***@***.***> wrote: Hi there, Thank you for your detailed report! It seems like you're encountering an issue with the libgl1-mesa-glx package on Ubuntu 24.04 LTS. This package is essential for OpenCV to function correctly, and its absence can cause the errors you're seeing. To resolve this, you can try installing the libgl1 package, which should provide the necessary OpenGL libraries. Please run the following commands in your terminal: sudo apt-get update sudo apt-get install -y libgl1 If the issue persists, it might be due to changes or updates in the package repositories for Ubuntu 24.04 LTS. In such cases, you can also try installing the libgl1-mesa-glx package from an older Ubuntu release or using a different package source. Additionally, please ensure that you are using the latest versions of torch and ultralytics. You can update them using: pip install --upgrade torch ultralytics Once you've installed the necessary libraries and updated your packages, try running your YOLOv8 command again: yolo task=detect mode=predict model=yolov8n.pt source='https://ultralytics.com/images/zidane.jpg' If you continue to experience issues, please provide a minimum reproducible code example so we can investigate further. You can find more information on creating a reproducible example here: Minimum Reproducible Example <https://docs.ultralytics.com/help/minimum_reproducible_example>. Feel free to reach out if you have any more questions or need further assistance! 😊 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQVJMDCAHS5HMBTC5ELZGNKJBAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMJVHEYTI> . You are receiving this because you commented.Message ID: ***@***.*** com>

[pderrenger]

@adiyazid hi there,

Thank you for reaching out and for resolving the initial issue! I'm glad to hear that you're ready to dive into using the ParkingPtsSelection tool. Let's walk through the process of defining parking areas and saving the configuration as a JSON file.

Step-by-Step Guide to Using ParkingPtsSelection

1. Capture a Frame:

   • First, capture a frame from the video or camera stream where you want to manage the parking lot. This frame will serve as the basis for defining parking regions.

2. Launch the ParkingPtsSelection Tool:

   • Use the following code snippet to launch the graphical interface for selecting parking points:

     from ultralytics import solutions
     
     solutions.ParkingPtsSelection()

   • This will open a window where you can select an image and start outlining parking regions by clicking to create polygons.

3. Define Parking Areas:

   • Click around the parking spaces to create polygons that define each parking area. You can adjust the points as needed to accurately outline the spaces.

4. Save the Configuration:

   • Once you have defined all the parking areas, click the save button. This will store the configuration as a JSON file in your working directory.

Example Code for Parking Management

Here’s a complete example of how to use the saved JSON file for parking management with YOLOv8:

import cv2
from ultralytics import solutions

# Path to the JSON file created with the ParkingPtsSelection tool
polygon_json_path = "bounding_boxes.json"

# Video capture
cap = cv2.VideoCapture("Path/to/video/file.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

# Video writer
video_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

# Initialize parking management object
management = solutions.ParkingManagement(model_path="yolov8n.pt")

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

Optional Arguments for ParkingManagement

You can customize the appearance and behavior of the parking management system using the following optional arguments:

Name	Type	Default	Description
model_path	str	None	Path to the YOLOv8 model.
txt_color	tuple	(0, 0, 0)	RGB color tuple for text.
bg_color	tuple	(255, 255, 255)	RGB color tuple for background.
occupied_region_color	tuple	(0, 255, 0)	RGB color tuple for occupied regions.
available_region_color	tuple	(0, 0, 255)	RGB color tuple for available regions.
margin	int	10	Margin for text display.

For more detailed information, you can refer to the Parking Management Guide.

Feel free to reach out if you have any more questions or need further assistance! 😊

[adiyazid]

1. I cannot launch the graphical interface using script -> python select_park.py The file contain the following code. from ultralytics import solutionssolutions.ParkingPtsSelection() 2. The error message as follows. requirements: Ultralytics requirement ['tkinter'] not found, attempting AutoUpdate... ERROR: Could not find a version that satisfies the requirement tkinter (from versions: none) ERROR: No matching distribution found for tkinter Retry 1/2 failed: Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1. ERROR: Could not find a version that satisfies the requirement tkinter (from versions: none) ERROR: No matching distribution found for tkinter Retry 2/2 failed: Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1. requirements: ❌ Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1. Traceback (most recent call last): File "/home/kpark/select_park.py", line 3, in <module> solutions.ParkingPtsSelection() File "/yolov8env/lib/python3.12/site-packages/ultralytics/solutions/parking_management.py", line 22, in __init__ self.master = tk.Tk() ^^^^^^^ File "/usr/lib/python3.12/tkinter/__init__.py", line 2345, in __init__ self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ _tkinter.TclError: no display name and no $DISPLAY environment variable Please assist on this, thanks.

…

On Sun, Jun 9, 2024 at 8:15 PM Paula Derrenger ***@***.***> wrote: @adiyazid <https://github.com/adiyazid> hi there, Thank you for reaching out and for resolving the initial issue! I'm glad to hear that you're ready to dive into using the ParkingPtsSelection tool. Let's walk through the process of defining parking areas and saving the configuration as a JSON file. Step-by-Step Guide to Using ParkingPtsSelection 1. *Capture a Frame:* - First, capture a frame from the video or camera stream where you want to manage the parking lot. This frame will serve as the basis for defining parking regions. 2. *Launch the ParkingPtsSelection Tool:* - Use the following code snippet to launch the graphical interface for selecting parking points: from ultralytics import solutions solutions.ParkingPtsSelection() - This will open a window where you can select an image and start outlining parking regions by clicking to create polygons. 3. *Define Parking Areas:* - Click around the parking spaces to create polygons that define each parking area. You can adjust the points as needed to accurately outline the spaces. 4. *Save the Configuration:* - Once you have defined all the parking areas, click the save button. This will store the configuration as a JSON file in your working directory. Example Code for Parking Management Here’s a complete example of how to use the saved JSON file for parking management with YOLOv8: import cv2from ultralytics import solutions # Path to the JSON file created with the ParkingPtsSelection toolpolygon_json_path = "bounding_boxes.json" # Video capturecap = cv2.VideoCapture("Path/to/video/file.mp4")assert cap.isOpened(), "Error reading video file"w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) # Video writervideo_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Initialize parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") while cap.isOpened(): ret, im0 = cap.read() if not ret: break json_data = management.parking_regions_extraction(polygon_json_path) results = management.model.track(im0, persist=True, show=False) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) management.display_frames(im0) video_writer.write(im0) cap.release()video_writer.release()cv2.destroyAllWindows() Optional Arguments for ParkingManagement You can customize the appearance and behavior of the parking management system using the following optional arguments: Name Type Default Description model_path str None Path to the YOLOv8 model. txt_color tuple (0, 0, 0) RGB color tuple for text. bg_color tuple (255, 255, 255) RGB color tuple for background. occupied_region_color tuple (0, 255, 0) RGB color tuple for occupied regions. available_region_color tuple (0, 0, 255) RGB color tuple for available regions. margin int 10 Margin for text display. For more detailed information, you can refer to the Parking Management Guide <https://docs.ultralytics.com/guides/parking-management/>. Feel free to reach out if you have any more questions or need further assistance! 😊 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQW5CJFRX7PVO2NZCN3ZGRBOTAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMJYHA3DQ> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[pderrenger]

Hi @adiyazid,

Thank you for reaching out and providing the detailed error message. It looks like you're encountering issues with launching the graphical interface due to missing tkinter and display environment variables.

Steps to Resolve the Issue:

1. Install tkinter:
   tkinter is a standard GUI library for Python, but it might not be installed by default in some environments. You can install it using the following command:

   sudo apt-get install python3-tk

   If you're using a different operating system, please refer to the appropriate package manager for installation.

2. Set Display Environment Variable:
   The error no display name and no $DISPLAY environment variable indicates that the script is trying to open a GUI window but can't find a display. This is common in headless environments like remote servers. If you're running this on a local machine with a GUI, ensure that your display environment variable is set correctly:

   export DISPLAY=:0

   If you're on a remote server, you might need to use X11 forwarding or a virtual display like Xvfb.

3. Verify Package Versions:
   Ensure that you're using the latest versions of torch and ultralytics. You can update them using:

   pip install --upgrade torch ultralytics

Example Code for Parking Management:

Here's a complete example of how to use the saved JSON file for parking management with YOLOv8:

import cv2
from ultralytics import solutions

# Path to json file, that created with above point selection app
polygon_json_path = "bounding_boxes.json"

# Video capture
cap = cv2.VideoCapture("Path/to/video/file.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

# Video writer
video_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

# Initialize parking management object
management = solutions.ParkingManagement(model_path="yolov8n.pt")

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

For more detailed information, you can refer to the Parking Management Guide.

If you continue to experience issues, please provide a minimum reproducible example of your code and ensure you are using the latest versions of the required packages. This will help us investigate and resolve the issue more effectively.

Feel free to reach out if you have any more questions or need further assistance! 😊

[adiyazid]

Hi, 1. I’m still having problem to launch it. Can I open it locally using Macbook Air M3? 2. My datasets are static parking bay images captured from camera via ftp every 1 minute stored in ftp server. Can I still use the same code for parking management?

…

On Mon, 10 Jun 2024 at 3:31 AM, Paula Derrenger ***@***.***> wrote: Hi @adiyazid <https://github.com/adiyazid>, Thank you for reaching out and providing the detailed error message. It looks like you're encountering issues with launching the graphical interface due to missing tkinter and display environment variables. Steps to Resolve the Issue: 1. *Install tkinter:* tkinter is a standard GUI library for Python, but it might not be installed by default in some environments. You can install it using the following command: sudo apt-get install python3-tk If you're using a different operating system, please refer to the appropriate package manager for installation. 2. *Set Display Environment Variable:* The error no display name and no $DISPLAY environment variable indicates that the script is trying to open a GUI window but can't find a display. This is common in headless environments like remote servers. If you're running this on a local machine with a GUI, ensure that your display environment variable is set correctly: export DISPLAY=:0 If you're on a remote server, you might need to use X11 forwarding or a virtual display like Xvfb. 3. *Verify Package Versions:* Ensure that you're using the latest versions of torch and ultralytics. You can update them using: pip install --upgrade torch ultralytics Example Code for Parking Management: Here's a complete example of how to use the saved JSON file for parking management with YOLOv8: import cv2from ultralytics import solutions # Path to json file, that created with above point selection apppolygon_json_path = "bounding_boxes.json" # Video capturecap = cv2.VideoCapture("Path/to/video/file.mp4")assert cap.isOpened(), "Error reading video file"w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) # Video writervideo_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Initialize parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") while cap.isOpened(): ret, im0 = cap.read() if not ret: break json_data = management.parking_regions_extraction(polygon_json_path) results = management.model.track(im0, persist=True, show=False) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) management.display_frames(im0) video_writer.write(im0) cap.release()video_writer.release()cv2.destroyAllWindows() For more detailed information, you can refer to the Parking Management Guide <https://docs.ultralytics.com/guides/parking-management/>. If you continue to experience issues, please provide a minimum reproducible example of your code and ensure you are using the latest versions of the required packages. This will help us investigate and resolve the issue more effectively. Feel free to reach out if you have any more questions or need further assistance! 😊 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQT3WQSK3BW4T2FU4STZGSUQRAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMRQG43DI> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[RizwanMunawar]

@adiyazid Thanks for your inquiry! While I can't confirm if the GUI will function on a MacBook, I can assure you it works on both Linux and Windows environments. You can use this code for parking management, provided that your camera is stationary.

Thanks,
The Ultralytics Team

[adiyazid]

1. I've managed to launch it with Windows10 and create the JSON file. 2. I notice that there is a code expecting a video stream -> cap = cv2. VideoCapture("Path/to/video/file.mp4"). I don't have any video stream, I have datasets of static images from ftp server capturing images every 1 minute from the camera. How do I handle this?

…

On Mon, Jun 10, 2024 at 8:51 PM Muhammad Rizwan Munawar < ***@***.***> wrote: @adiyazid <https://github.com/adiyazid> Thanks for your inquiry! While I can't confirm if the GUI will function on a MacBook, I can assure you it works on both Linux and Windows environments. You can use this code for parking management, provided that your camera is stationary. Thanks, The Ultralytics Team — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQWR7LIAFRQDCZUG4PDZGWOOHAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMRXGA3TC> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[pderrenger]

Hi @adiyazid,

Thank you for your detailed question! It's great to hear that you've successfully created the JSON file on Windows 10. Let's address your concern about handling static images instead of a video stream.

You can modify the provided code to work with a dataset of static images by iterating through your image files instead of reading frames from a video stream. Here's an example of how you can adapt the code to process images from a directory:

import cv2
import os
from ultralytics import solutions

# Path to the JSON file created with the point selection app
polygon_json_path = "bounding_boxes.json"

# Directory containing your static images
image_dir = "path/to/your/image/directory"

# Initialize the parking management object
management = solutions.ParkingManagement(model_path="yolov8n.pt")

# Get a list of all image files in the directory
image_files = [os.path.join(image_dir, f) for f in os.listdir(image_dir) if f.endswith(('.png', '.jpg', '.jpeg'))]

# Loop through each image file
for image_file in image_files:
    # Read the image
    im0 = cv2.imread(image_file)
    if im0 is None:
        print(f"Error reading image {image_file}")
        continue

    # Extract parking regions from the JSON data
    json_data = management.parking_regions_extraction(polygon_json_path)
    
    # Perform tracking on the image
    results = management.model.track(im0, persist=True, show=False)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        management.process_data(json_data, im0, boxes, clss)

    # Display the processed image
    management.display_frames(im0)
    cv2.imshow("Parking Management", im0)
    cv2.waitKey(1)  # Display each image for 1 ms

cv2.destroyAllWindows()

In this code:

1. We read images from a specified directory.
2. For each image, we perform the same processing as we would for a video frame.
3. The results are displayed using OpenCV's imshow.

This approach should allow you to handle static images captured at regular intervals. If you have any further questions or run into issues, feel free to ask!

[adiyazid]

Hi, How about fisheye camera? Will there be any issue to train the dataset?

…

On Tue, 11 Jun 2024 at 4:22 AM, Paula Derrenger ***@***.***> wrote: Hi @adiyazid <https://github.com/adiyazid>, Thank you for your detailed question! It's great to hear that you've successfully created the JSON file on Windows 10. Let's address your concern about handling static images instead of a video stream. You can modify the provided code to work with a dataset of static images by iterating through your image files instead of reading frames from a video stream. Here's an example of how you can adapt the code to process images from a directory: import cv2import osfrom ultralytics import solutions # Path to the JSON file created with the point selection apppolygon_json_path = "bounding_boxes.json" # Directory containing your static imagesimage_dir = "path/to/your/image/directory" # Initialize the parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") # Get a list of all image files in the directoryimage_files = [os.path.join(image_dir, f) for f in os.listdir(image_dir) if f.endswith(('.png', '.jpg', '.jpeg'))] # Loop through each image filefor image_file in image_files: # Read the image im0 = cv2.imread(image_file) if im0 is None: print(f"Error reading image {image_file}") continue # Extract parking regions from the JSON data json_data = management.parking_regions_extraction(polygon_json_path) # Perform tracking on the image results = management.model.track(im0, persist=True, show=False) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) # Display the processed image management.display_frames(im0) cv2.imshow("Parking Management", im0) cv2.waitKey(1) # Display each image for 1 ms cv2.destroyAllWindows() In this code: 1. We read images from a specified directory. 2. For each image, we perform the same processing as we would for a video frame. 3. The results are displayed using OpenCV's imshow. This approach should allow you to handle static images captured at regular intervals. If you have any further questions or run into issues, feel free to ask! — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQXFT7CST22E4TQSNIDZGYDHPAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMZRGQ2DK> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[pderrenger]

Hi @adiyazid,

Thank you for your question! Using a fisheye camera can indeed present some unique challenges due to the distortion inherent in fisheye lenses. However, YOLOv8 is quite robust and can handle various types of input data, including fisheye images, with some preprocessing.

Key Considerations:

1. Preprocessing: You might need to undistort the fisheye images before feeding them into the YOLOv8 model. OpenCV provides tools for fisheye image correction which can be very useful.
2. Training Data: Ensure that your training dataset includes fisheye images or images that have been preprocessed to simulate the fisheye effect. This will help the model learn to recognize objects in such images more effectively.

Example Preprocessing Code:

Here's a simple example using OpenCV to undistort fisheye images:

import cv2
import numpy as np

# Load the fisheye image
img = cv2.imread('fisheye_image.jpg')

# Define the camera matrix and distortion coefficients
K = np.array([[fx, 0, cx],
              [0, fy, cy],
              [0, 0, 1]])
D = np.array([k1, k2, k3, k4])

# Undistort the image
h, w = img.shape[:2]
new_K = cv2.fisheye.estimateNewCameraMatrixForUndistortRectify(K, D, (w, h), np.eye(3), balance=1)
map1, map2 = cv2.fisheye.initUndistortRectifyMap(K, D, np.eye(3), new_K, (w, h), cv2.CV_16SC2)
undistorted_img = cv2.remap(img, map1, map2, interpolation=cv2.INTER_LINEAR, borderMode=cv2.BORDER_CONSTANT)

# Save or display the undistorted image
cv2.imwrite('undistorted_image.jpg', undistorted_img)

Training with YOLOv8:

Once you have your undistorted images, you can proceed with training YOLOv8 as usual. If you encounter any issues, please ensure you are using the latest versions of torch and ultralytics. You can update your packages using:

pip install --upgrade torch ultralytics

For more detailed guidance on parking management using YOLOv8, you can refer to our Parking Management Guide.

If you have any further questions or run into issues, feel free to ask! 😊

[prasadjaywant]

When tried to execute {from ultralytics import solutions
solutions.ParkingPtsSelection()} , getting following messages;

[
requirements: Ultralytics requirement ['tkinter'] not found, attempting AutoUpdate...
ERROR: Could not find a version that satisfies the requirement tkinter (from versions: none)
ERROR: No matching distribution found for tkinter
Retry 1/2 failed: Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1.
ERROR: Could not find a version that satisfies the requirement tkinter (from versions: none)
ERROR: No matching distribution found for tkinter
Retry 2/2 failed: Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1.
requirements: ❌ Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1.]

Please can you help anything needed to install tkinter? Thanks for your attention!

[glenn-jocher]

Hello @prasadjaywant,

Thank you for reaching out and providing detailed information about your use case and the issues you're encountering. Let's address your concerns step-by-step.

1. Expected Results for Vehicle and Person Detection

To ensure that vehicles and people are correctly identified, please verify that you are using the latest versions of torch and ultralytics. You can update them using:

pip install --upgrade torch ultralytics

2. Incorrect Class Labels

To filter out irrelevant classes and focus on specific ones like 'Car', 'Van', 'Truck', 'Person', etc., you can use the classes parameter in the track method. Here’s an example:

results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 15, 16, 17])

Replace the class indices with the ones relevant to your use case.

3. Customizing the Model for Specific Classes

If you want to train the model to recognize only specific classes, you can fine-tune the model on a custom dataset that includes only those classes. This will help in reducing the number of epochs and improving efficiency. You can create a custom dataset and train the model as follows:

from ultralytics import YOLO

# Load a model
model = YOLO('yolov8n-seg.pt')  # Load a pretrained model

# Train the model on your custom dataset
model.train(data='path/to/custom_dataset.yaml', epochs=50, imgsz=640)

Ensure your custom dataset YAML file is properly configured with the relevant classes.

4. Handling Incorrect Identifications

To handle incorrect identifications, you can adjust the confidence threshold and IoU threshold parameters. Increasing the confidence threshold can help reduce false positives:

results = management.model.track(im0, persist=True, show=False, conf=0.5, iou=0.6)

Example Code

Here’s a revised version of your code snippet incorporating the above suggestions:

import cv2
from ultralytics import solutions

# Path to json file created with an external tool
polygon_json_path = "bounding_boxes.json"

# Video capture
cap = cv2.VideoCapture("Path/to/video/file.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

# Video writer
video_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

# Initialize parking management object
management = solutions.ParkingManagement(model_path="yolov8n-seg.pt")

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 15, 16, 17], conf=0.5, iou=0.6)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

For more detailed guidance, please refer to our Parking Management Guide.

Feel free to reach out if you have any further questions or need additional assistance. We're here to help! 😊

[prasadjaywant]

Thanks @glenn-jocher and @[pderrenger] for your quick help. My further working is summarized as follows.
Observations:

1. Parking Annotator persists the 'tkinter not found error'. But surprisingly UI widget opens up, it accepts image for upload, allows polygonic annotations and creates bounding_boxes.json successfully as enclosed below.
2. For annotation image was created out of one of the frames in the video.
3. All code runs successfully. It's executed with pretrained model as-is. No customization with any new dataset is done yet.
4. .avi is generated.
5. Seemingly followed all steps and also the pip install --upgrade torch ultralytics.
6. However, no single object detection/segmentation label is generated during the processing and final .avi. The Python code execution was performed on local CPU and also in Google Colab. Same behavior is observed on both.
7. Are we missing something or anything not in order? The case was tried with 1 more video https://videos.pexels.com/video-files/3858833/3858833-hd_1280_720_24fps.mp4 and relevant bounding_boxes points. Result is same.

---

bounding_boxes_5587732-hd_1280_720_30fps.json:

[
{
"points": [
[
293,
129
],
[
375,
128
],
[
375,
159
],
[
293,
161
]
]
},
{
"points": [
[
291,
195
],
[
373,
195
],
[
373,
225
],
[
290,
225
]
]
},
{
"points": [
[
1214,
144
],
[
1291,
140
],
[
1293,
170
],
[
1209,
173
]
]
},
{
"points": [
[
1315,
367
],
[
1384,
368
],
[
1383,
397
],
[
1315,
395
]
]
},
{
"points": [
[
1297,
591
],
[
1368,
591
],
[
1368,
622
],
[
1296,
619
]
]
},
{
"points": [
[
1770,
368
],
[
1837,
367
],
[
1839,
392
],
[
1768,
394
]
]
},
{
"points": [
[
946,
442
],
[
980,
440
],
[
983,
504
],
[
948,
500
]
]
}
]

---

import cv2
from ultralytics import solutions

Path to json file created with an external tool

polygon_json_path = "bounding_boxes_5587732-hd_1280_720_30fps.json"

Video capture

cap = cv2.VideoCapture("https://videos.pexels.com/video-files/5587732/5587732-hd_1280_720_30fps.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

Video writer

video_writer = cv2.VideoWriter("parking_management_5587732-hd_1280_720_30fps.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

Initialize parking management object

management = solutions.ParkingManagement(model_path="yolov8n-seg.pt")
'''
management = solutions.ParkingManagement(
model_path="yolov8n-seg.pt",
txt_color=(0, 0, 0),
bg_color=(255, 255, 255),
occupied_region_color=(0, 255, 0),
available_region_color=(0, 0, 255),
margin=10
)
'''
while cap.isOpened():
ret, im0 = cap.read()
if not ret:
break

json_data = management.parking_regions_extraction(polygon_json_path)
results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 7], conf=0.5, iou=0.6)

classes=[0-person, 1-bicycle, 2-car, 3-motorcycle, 7-truck, 15-cat, 16-dog, 17-horse, 26-handbag, 28-suitcase]

if results[0].boxes.id is not None:
    boxes = results[0].boxes.xyxy.cpu().tolist()
    clss = results[0].boxes.cls.cpu().tolist()
    management.process_data(json_data, im0, boxes, clss)

management.display_frames(im0)
video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

---

[glenn-jocher]

Hello @prasadjaywant,

Thank you for your detailed summary and for providing the JSON file and code snippet. Let's address the issues you're encountering:

Observations and Issues:

1. tkinter Error: It's good to know that despite the error, the UI widget opens and functions correctly. This might be a minor issue with the environment setup but doesn't seem to affect the functionality.
2. No Object Detection/Segmentation Labels: The main issue appears to be that no objects are being detected or segmented in your video.

Steps to Resolve:

1. Verify Model and Dependencies:
   Ensure you are using the latest versions of torch and ultralytics. You can update them using:

   pip install --upgrade torch ultralytics

2. Check Class Labels:
   Ensure that the class labels you are filtering for are correct. The COCO dataset class indices are as follows:

   • 0: 'person'
   • 1: 'bicycle'
   • 2: 'car'
   • 3: 'motorcycle'
   • 7: 'truck'
   • 15: 'cat'
   • 16: 'dog'
   • 17: 'horse'
   • 26: 'handbag'
   • 28: 'suitcase'

   Make sure these indices match the classes you expect in your video.

3. Adjust Confidence and IoU Thresholds:
   Sometimes, adjusting the confidence and IoU thresholds can help in detecting objects more accurately. You can try lowering the confidence threshold to see if it helps:

   results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 7], conf=0.3, iou=0.5)

4. Debugging:
   Add some print statements to check if any detections are being made:

   if results[0].boxes.id is not None:
       boxes = results[0].boxes.xyxy.cpu().tolist()
       clss = results[0].boxes.cls.cpu().tolist()
       print(f"Detected boxes: {boxes}")
       print(f"Detected classes: {clss}")
       management.process_data(json_data, im0, boxes, clss)

Example Code:

Here’s the revised code snippet incorporating the above suggestions:

import cv2
from ultralytics import solutions

# Path to json file created with an external tool
polygon_json_path = "bounding_boxes_5587732-hd_1280_720_30fps.json"
# Video capture
cap = cv2.VideoCapture("https://videos.pexels.com/video-files/5587732/5587732-hd_1280_720_30fps.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

# Video writer
video_writer = cv2.VideoWriter("parking_management_5587732-hd_1280_720_30fps.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

# Initialize parking management object
management = solutions.ParkingManagement(model_path="yolov8n-seg.pt")

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 7], conf=0.3, iou=0.5)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        print(f"Detected boxes: {boxes}")
        print(f"Detected classes: {clss}")
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

Additional Resources:

For more detailed guidance, please refer to our Parking Management Guide.

Feel free to reach out if you have any further questions or need additional assistance. We're here to help! 😊

[prasadjaywant]

Dear @glenn-jocher and @pederrenger,

Terminal is showing preprocessing and postprocessing execution with class names as 'car', 'person' etc. but at very low scores of conf=0.05 and iou=0.1. The .avi doesn't show any detection label, box, mask or segment etc. for any of the objects. It just forms a vanilla clip same as source video. Also no execution of print() statements seen in the terminal. Any clue for rendering the annotations properly? Thanks very much......

[glenn-jocher]

Hello @prasadjaywant,

Thank you for your detailed report and for providing the context. Let's address the issue you're encountering with the annotations not rendering properly in the .avi file.

Steps to Resolve:

1. Verify Package Versions:
   Ensure you are using the latest versions of torch and ultralytics. You can update them using:

   pip install --upgrade torch ultralytics

2. Adjust Confidence and IoU Thresholds:
   Since you mentioned that the terminal shows low confidence scores, try adjusting the confidence and IoU thresholds to higher values to filter out low-confidence detections:

   results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 7], conf=0.5, iou=0.6)

3. Check for Detections:
   Add print statements to check if any detections are being made and to ensure the results are being processed correctly:

   if results[0].boxes.id is not None:
       boxes = results[0].boxes.xyxy.cpu().tolist()
       clss = results[0].boxes.cls.cpu().tolist()
       print(f"Detected boxes: {boxes}")
       print(f"Detected classes: {clss}")
       management.process_data(json_data, im0, boxes, clss)
   else:
       print("No detections found.")

4. Rendering Annotations:
   Ensure that the display_frames method is correctly rendering the annotations. You can also try displaying the frames directly to verify:

   management.display_frames(im0)
   cv2.imshow("Frame", im0)
   cv2.waitKey(1)

Example Code:

Here’s the revised code snippet incorporating the above suggestions:

import cv2
from ultralytics import solutions

# Path to json file created with an external tool
polygon_json_path = "bounding_boxes_5587732-hd_1280_720_30fps.json"
# Video capture
cap = cv2.VideoCapture("https://videos.pexels.com/video-files/5587732/5587732-hd_1280_720_30fps.mp4")
assert cap.isOpened(), "Error reading video file"
w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))

# Video writer
video_writer = cv2.VideoWriter("parking_management_5587732-hd_1280_720_30fps.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))

# Initialize parking management object
management = solutions.ParkingManagement(model_path="yolov8n-seg.pt")

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 7], conf=0.5, iou=0.6)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        print(f"Detected boxes: {boxes}")
        print(f"Detected classes: {clss}")
        management.process_data(json_data, im0, boxes, clss)
    else:
        print("No detections found.")

    management.display_frames(im0)
    cv2.imshow("Frame", im0)  # Display the frame to verify annotations
    cv2.waitKey(1)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

Additional Resources:

For more detailed guidance, please refer to our Parking Management Guide.

Feel free to reach out if you have any further questions or need additional assistance. We're here to help! 😊

[adiyazid]

Hi, I've decided to use Windows to run ParkingPtsSelection and it works fine. Do you provide development service if I need to? What's your rate? Do let me know, thanks.

…

On Fri, Jun 14, 2024 at 6:16 AM Glenn Jocher ***@***.***> wrote: @prasadjaywant <https://github.com/prasadjaywant> hello! Thank you for reaching out and providing detailed information about your setup and the issue you're encountering. It appears that the tkinter library is not being recognized by your Python environment, which is necessary for the ParkingPtsSelection tool to function. Steps to Resolve the Issue: 1. *Verify tkinter Installation:* Since tkinter is included with standard Python installations, it should be available. However, it might not be installed by default in some environments. You can try installing it manually: sudo apt-get install python3-tk For Windows, tkinter should be included with the Python installer. Ensure that your Python installation is complete and includes all optional components. 2. *Check Python Path:* Ensure that your Python environment is correctly set up and that the tkinter library is in the correct path. You can verify this by running: import tkinterprint(tkinter.__file__) This should print the path to the tkinter module. If it doesn't, there might be an issue with your Python installation. 3. *Alternative Annotation Tools:* If you continue to face issues with tkinter, you can use external tools like Roboflow to create polygonal annotations. Once you have created the annotations, save them as bounding_boxes.json and proceed with the parking management workflow. Example Workflow Without ParkingPtsSelection: 1. *Create Annotations Externally:* Use an external tool to create polygonal annotations and save them as bounding_boxes.json. 2. *Run Parking Management:* Use the following code to execute the parking management system: import cv2from ultralytics import solutions # Path to json file created with an external toolpolygon_json_path = "bounding_boxes.json" # Video capturecap = cv2.VideoCapture("Path/to/video/file.mp4")assert cap.isOpened(), "Error reading video file"w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) # Video writervideo_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Initialize parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") while cap.isOpened(): ret, im0 = cap.read() if not ret: break json_data = management.parking_regions_extraction(polygon_json_path) results = management.model.track(im0, persist=True, show=False) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) management.display_frames(im0) video_writer.write(im0) cap.release()video_writer.release()cv2.destroyAllWindows() For more detailed guidance, please refer to our Parking Management Guide <https://docs.ultralytics.com/guides/parking-management/>. If you continue to experience issues, please ensure you are using the latest versions of torch and ultralytics. You can update them using: pip install --upgrade torch ultralytics Feel free to reach out if you have any further questions or need additional assistance. We're here to help! 😊 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQTDFFBMSE6KJ3VQ4C3ZHIK23AVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TONRYGE4DM> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[prasadjaywant]

Thank you Glenn for your prompt revert. Will work out the given steps and come back soon.....

…

On Wed, 19 Jun 2024, 5:35 am Glenn Jocher, ***@***.***> wrote: Hello @prasadjaywant <https://github.com/prasadjaywant>, Thank you for reaching out and providing detailed information about your use case and the issues you're encountering. Let's address your concerns step-by-step. 1. Expected Results for Vehicle and Person Detection To ensure that vehicles and people are correctly identified, please verify that you are using the latest versions of torch and ultralytics. You can update them using: pip install --upgrade torch ultralytics 2. Incorrect Class Labels To filter out irrelevant classes and focus on specific ones like 'Car', 'Van', 'Truck', 'Person', etc., you can use the classes parameter in the track method. Here’s an example: results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 15, 16, 17]) Replace the class indices with the ones relevant to your use case. 3. Customizing the Model for Specific Classes If you want to train the model to recognize only specific classes, you can fine-tune the model on a custom dataset that includes only those classes. This will help in reducing the number of epochs and improving efficiency. You can create a custom dataset and train the model as follows: from ultralytics import YOLO # Load a modelmodel = YOLO('yolov8n-seg.pt') # Load a pretrained model # Train the model on your custom datasetmodel.train(data='path/to/custom_dataset.yaml', epochs=50, imgsz=640) Ensure your custom dataset YAML file is properly configured with the relevant classes. 4. Handling Incorrect Identifications To handle incorrect identifications, you can adjust the confidence threshold and IoU threshold parameters. Increasing the confidence threshold can help reduce false positives: results = management.model.track(im0, persist=True, show=False, conf=0.5, iou=0.6) Example Code Here’s a revised version of your code snippet incorporating the above suggestions: import cv2from ultralytics import solutions # Path to json file created with an external toolpolygon_json_path = "bounding_boxes.json" # Video capturecap = cv2.VideoCapture("Path/to/video/file.mp4")assert cap.isOpened(), "Error reading video file"w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS)) # Video writervideo_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h)) # Initialize parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n-seg.pt") while cap.isOpened(): ret, im0 = cap.read() if not ret: break json_data = management.parking_regions_extraction(polygon_json_path) results = management.model.track(im0, persist=True, show=False, classes=[0, 1, 2, 3, 15, 16, 17], conf=0.5, iou=0.6) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) management.display_frames(im0) video_writer.write(im0) cap.release()video_writer.release()cv2.destroyAllWindows() For more detailed guidance, please refer to our Parking Management Guide <https://docs.ultralytics.com/guides/parking-management/>. Feel free to reach out if you have any further questions or need additional assistance. We're here to help! 😊 — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQL2CZCHSTGNNGIEHQJE7SLZIDDLTAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TQMJRGU2DA> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[pderrenger]

You're welcome, @prasadjaywant! 😊

I'm glad to hear you're working through the steps. If you encounter any issues or need further assistance, please don't hesitate to reach out.

Just a quick reminder:

• Ensure you're using the latest versions of torch and ultralytics by running:

  pip install --upgrade torch ultralytics

• If you run into any bugs, please provide a minimum reproducible code example. This helps us investigate and resolve the issue more efficiently. You can find more details on creating a reproducible example here.

For more detailed guidance on parking management, you can refer to our Parking Management Guide.

Looking forward to your updates! 🚀

[charan112]

I'm Using Python IDLE, The Python Program is working but The recognition is not idle and the values of occupancy and available is not changing accordingly. And the the cars are recognized as cell phones.

[pderrenger]

@charan112 hi there! 😊

Thank you for reaching out. It sounds like you're experiencing a couple of issues with your parking management system using YOLOv8. Let's address them one by one.

1. Recognition Issues

If cars are being recognized as cell phones, it might be due to the model not being trained on the appropriate dataset or the confidence threshold settings. Here are a few steps to troubleshoot:

1. Ensure Correct Model: Verify that you are using a model trained specifically for vehicle detection. The yolov8n.pt model might not be trained on a dataset that includes cars. You might need a custom-trained model on a relevant dataset like COCO or a specific vehicle dataset.

2. Adjust Confidence Threshold: Sometimes, adjusting the confidence threshold can help improve detection accuracy. You can set the conf parameter in the track method to a higher value, such as 0.5.

   results = management.model.track(im0, persist=True, show=False, conf=0.5)

3. Filter Classes: Ensure that the model is set to detect the correct classes. You can filter the results by class using the classes parameter. For example, if cars are class 2 in your model:

   results = management.model.track(im0, persist=True, show=False, classes=[2])

2. Occupancy and Availability Values Not Updating

This could be due to the logic in the process_data method or the way the regions are defined. Here are a few steps to check:

1. Verify JSON Data: Ensure that the JSON file created with the ParkingPtsSelection tool correctly defines the parking regions.

2. Check process_data Logic: Make sure that the process_data method correctly updates the occupancy and availability based on the detected boxes and classes.

3. Debugging: Add print statements or use a debugger to check the values of boxes and clss to ensure they are being correctly passed to the process_data method.

Example Code for Debugging

Here's an example of how you might add some debugging to your loop:

while cap.isOpened():
    ret, im0 = cap.read()
    if not ret:
        break

    json_data = management.parking_regions_extraction(polygon_json_path)
    results = management.model.track(im0, persist=True, show=False)

    if results[0].boxes.id is not None:
        boxes = results[0].boxes.xyxy.cpu().tolist()
        clss = results[0].boxes.cls.cpu().tolist()
        
        # Debugging: Print detected boxes and classes
        print(f"Detected boxes: {boxes}")
        print(f"Detected classes: {clss}")
        
        management.process_data(json_data, im0, boxes, clss)

    management.display_frames(im0)
    video_writer.write(im0)

cap.release()
video_writer.release()
cv2.destroyAllWindows()

Additional Resources

For more detailed guidance on setting up and troubleshooting your parking management system, please refer to our Parking Management Guide.

If the issue persists, please ensure you are using the latest versions of torch and ultralytics. You can update them using:

pip install --upgrade torch ultralytics

Feel free to share a minimum reproducible example if the problem continues, and we can dive deeper into the issue. You can find more information on creating a reproducible example here.

Best of luck with your project! 🚀

[prasadjaywant]

We are looking for possibility of adding some video look n feel cosmetic changes, also some custom business logic etc in the process_data method function of solutions.ParkingManagement class. Request to kindly guide me for any such possibility of customization in the Parking Management solution. Thanks n Regards......

[pderrenger]

Hi @prasadjaywant,

Thank you for your interest in customizing the Parking Management solution using Ultralytics YOLOv8! 😊

Customizing process_data Method

You can indeed add custom business logic and cosmetic changes to the process_data method of the solutions.ParkingManagement class. This method is designed to be flexible and can be modified to suit your specific requirements.

Steps for Customization

1. Subclassing ParkingManagement:
   Create a subclass of ParkingManagement and override the process_data method to include your custom logic.

   from ultralytics import solutions
   
   class CustomParkingManagement(solutions.ParkingManagement):
       def process_data(self, json_data, im0, boxes, clss):
           # Add your custom business logic here
           super().process_data(json_data, im0, boxes, clss)
           # Example: Custom cosmetic changes
           for box in boxes:
               x1, y1, x2, y2 = map(int, box)
               cv2.rectangle(im0, (x1, y1), (x2, y2), (255, 0, 0), 2)  # Blue rectangle
               cv2.putText(im0, "Custom Text", (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 2)

2. Using the Custom Class:
   Replace the instantiation of ParkingManagement with your custom class.

   import cv2
   
   # Path to json file, that created with above point selection app
   polygon_json_path = "bounding_boxes.json"
   
   # Video capture
   cap = cv2.VideoCapture("Path/to/video/file.mp4")
   assert cap.isOpened(), "Error reading video file"
   w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))
   
   # Video writer
   video_writer = cv2.VideoWriter("parking_management.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))
   
   # Initialize custom parking management object
   management = CustomParkingManagement(model_path="yolov8n.pt")
   
   while cap.isOpened():
       ret, im0 = cap.read()
       if not ret:
           break
   
       json_data = management.parking_regions_extraction(polygon_json_path)
       results = management.model.track(im0, persist=True, show=False)
   
       if results[0].boxes.id is not None:
           boxes = results[0].boxes.xyxy.cpu().tolist()
           clss = results[0].boxes.cls.cpu().tolist()
           management.process_data(json_data, im0, boxes, clss)
   
       management.display_frames(im0)
       video_writer.write(im0)
   
   cap.release()
   video_writer.release()
   cv2.destroyAllWindows()

Additional Tips

• Verify Model and Classes: Ensure that the model you are using is trained on the appropriate dataset for vehicle detection. Adjust the classes parameter in the track method to filter for vehicle classes.
• Debugging: Use print statements or a debugger to verify that the boxes and clss variables contain the expected values.

For more detailed guidance, please refer to our Parking Management Guide.

Feel free to reach out if you have any further questions. Happy coding! 🚀

[prasadjaywant]

Dear Paula @pderrenger,
Greetings.
Thanks for your quick revert. It opens up an opportunity to explore how best we could benefit from the derived class. Will research and revert if any limitations or changes could be useful to all.
Thanking you,
Regards,
Prasad

[pderrenger]

@prasadjaywant hi Prasad,

Thank you for your thoughtful response! 😊

I'm glad to hear that the derived class approach opens up new possibilities for your project. Customizing the process_data method and adding cosmetic changes can indeed enhance the functionality and user experience of your parking management system.

If you encounter any limitations or have suggestions for improvements that could benefit the broader community, please feel free to share them. Your feedback is invaluable in helping us refine and enhance our solutions.

For further customization, you might find the following resources helpful:

• Parking Management Guide: This guide provides detailed instructions and examples for setting up and customizing your parking management system.
• Minimum Reproducible Example: If you run into any issues, providing a reproducible example can help us diagnose and resolve the problem more efficiently.

Should you need to verify that the issue persists in the latest versions of the packages, please ensure you are using the most recent versions of torch and ultralytics. You can update them using:

pip install --upgrade torch ultralytics

Feel free to reach out if you have any further questions or need additional assistance. We're here to help!

[qbadat]

hi,
i have changed video file to a camera rtsp stream. Two issue i am finding when getting the results in real time.

1. there is like 40/50sec lag in seeing the tracking results
2. accuracy is not very good as even when parking slots are occupied, it reports as free.

Please help with the above two items. Thankyou

[glenn-jocher]

@qbadat hi there,

Thank you for reaching out! Let's address the two issues you're experiencing with your RTSP stream setup:

1. Lag in Tracking Results

The lag you're experiencing could be due to several factors, including network latency, processing time, or buffering issues. Here are a few steps to help mitigate this:

• Reduce Frame Rate: Lower the frame rate of your RTSP stream to reduce the processing load.
• Optimize Model Inference: Ensure you're using the latest version of YOLOv8 and consider exporting your model to a more optimized format like TensorRT for faster inference. You can follow the Export Guide for detailed instructions.
• Buffer Management: Adjust the buffer settings in your video capture code to minimize latency. For example, you can set the buffer size to a smaller value using OpenCV:

  cap.set(cv2.CAP_PROP_BUFFERSIZE, 1)

2. Accuracy of Parking Slot Detection

To improve the accuracy of detecting occupied and free parking slots, consider the following:

• Model Confidence and IOU Thresholds: Fine-tune the conf (confidence threshold) and iou (Intersection Over Union threshold) parameters in your tracking code. Higher thresholds can reduce false positives.

  results = management.model.track(im0, persist=True, conf=0.5, iou=0.6, show=False)

• Polygon Definition: Ensure that the polygons defining your parking slots are accurately drawn. Any inaccuracies in the polygon coordinates can lead to incorrect occupancy detection.
• Model Training: If you're using a pre-trained model, consider fine-tuning it on a dataset specific to your parking lot to improve detection accuracy.

Here's an example of how you might adjust the tracking parameters:

results = management.model.track(im0, persist=True, conf=0.5, iou=0.6, show=False)

Additionally, please verify that you are using the latest versions of all relevant packages to ensure you benefit from the latest improvements and bug fixes.

For more detailed guidance, you can refer to our Parking Management Guide.

I hope this helps! If you have any further questions, feel free to ask. 😊

[qbadat]

Thankyou for quick response, I’ll test out below suggestion and let you know. Btw, I am assuming “pip install ultralytics“ installs latest version along with its dependency. Thankyou

…

Sent from my iPhone On Jun 26, 2024, at 9:30 PM, Glenn Jocher ***@***.***> wrote:  @qbadat<https://github.com/qbadat> hi there, Thank you for reaching out! Let's address the two issues you're experiencing with your RTSP stream setup: 1. Lag in Tracking Results The lag you're experiencing could be due to several factors, including network latency, processing time, or buffering issues. Here are a few steps to help mitigate this: * Reduce Frame Rate: Lower the frame rate of your RTSP stream to reduce the processing load. * Optimize Model Inference: Ensure you're using the latest version of YOLOv8 and consider exporting your model to a more optimized format like TensorRT for faster inference. You can follow the Export Guide<https://docs.ultralytics.com/modes/export> for detailed instructions. * Buffer Management: Adjust the buffer settings in your video capture code to minimize latency. For example, you can set the buffer size to a smaller value using OpenCV: cap.set(cv2.CAP_PROP_BUFFERSIZE, 1) 2. Accuracy of Parking Slot Detection To improve the accuracy of detecting occupied and free parking slots, consider the following: * Model Confidence and IOU Thresholds: Fine-tune the conf (confidence threshold) and iou (Intersection Over Union threshold) parameters in your tracking code. Higher thresholds can reduce false positives. results = management.model.track(im0, persist=True, conf=0.5, iou=0.6, show=False) * Polygon Definition: Ensure that the polygons defining your parking slots are accurately drawn. Any inaccuracies in the polygon coordinates can lead to incorrect occupancy detection. * Model Training: If you're using a pre-trained model, consider fine-tuning it on a dataset specific to your parking lot to improve detection accuracy. Here's an example of how you might adjust the tracking parameters: results = management.model.track(im0, persist=True, conf=0.5, iou=0.6, show=False) Additionally, please verify that you are using the latest versions of all relevant packages to ensure you benefit from the latest improvements and bug fixes. For more detailed guidance, you can refer to our Parking Management Guide<https://docs.ultralytics.com/guides/parking-management/>. I hope this helps! If you have any further questions, feel free to ask. 😊 — Reply to this email directly, view it on GitHub<#11093 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AMWRSFUFS2CKAS4SRSAKEL3ZJNTKZAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TQOBYGU3TI>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[glenn-jocher]

@qbadat you're welcome! 😊 Yes, running pip install ultralytics will indeed install the latest version of Ultralytics YOLO along with its dependencies. This ensures you have the most up-to-date features and fixes.

If you encounter any issues, please make sure to provide a minimum reproducible example. This helps us diagnose and resolve the problem more efficiently. You can find guidance on creating one here: Minimum Reproducible Example.

Additionally, if you haven't already, please verify that the issue persists with the latest versions of all relevant packages.

Looking forward to hearing how your tests go! If you have any further questions or need additional assistance, feel free to ask. 🚀

[Izzatullokh24]

Assalamu alaykum everyone!
I am facing issue when using

from ultralytics import solutions

solutions.ParkingPtsSelection()

requirements: Ultralytics requirement ['tkinter'] not found, attempting AutoUpdate...
ERROR: Could not find a version that satisfies the requirement tkinter (from versions: none)
ERROR: No matching distribution found for tkinter
Retry 1/2 failed: Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1.
ERROR: Could not find a version that satisfies the requirement tkinter (from versions: none)
ERROR: No matching distribution found for tkinter
Retry 2/2 failed: Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1.
requirements: ❌ Command 'pip install --no-cache-dir "tkinter" ' returned non-zero exit status 1.

I am getting this error, but interface running and I can outline parking regions

I have checked tkinter library on my laptop, it has already installed(8.6 version). I am using windows 11.

[RizwanMunawar]

@Izzatullokh24 Hi there, this is just a warning and the code should be work fine. I would recommend start creation of parking zones and later you can simply save them in Json format for processing.

Thanks, Muhammad!
Ultralytics

[adiyazid]

Hi Paula, I've managed to use the platform for detection parking, at this moment I need to improve the accuracy Yolov8 does detect the vehicle but the results for parking availability is only based on cars, please find attached image. How do I change the code so that it improves the detection and detects an occupied bay regardless of vehicle. Please reply to me, soon, thank you.

…

On Tue, Jun 11, 2024 at 4:22 AM Paula Derrenger ***@***.***> wrote: Hi @adiyazid <https://github.com/adiyazid>, Thank you for your detailed question! It's great to hear that you've successfully created the JSON file on Windows 10. Let's address your concern about handling static images instead of a video stream. You can modify the provided code to work with a dataset of static images by iterating through your image files instead of reading frames from a video stream. Here's an example of how you can adapt the code to process images from a directory: import cv2import osfrom ultralytics import solutions # Path to the JSON file created with the point selection apppolygon_json_path = "bounding_boxes.json" # Directory containing your static imagesimage_dir = "path/to/your/image/directory" # Initialize the parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") # Get a list of all image files in the directoryimage_files = [os.path.join(image_dir, f) for f in os.listdir(image_dir) if f.endswith(('.png', '.jpg', '.jpeg'))] # Loop through each image filefor image_file in image_files: # Read the image im0 = cv2.imread(image_file) if im0 is None: print(f"Error reading image {image_file}") continue # Extract parking regions from the JSON data json_data = management.parking_regions_extraction(polygon_json_path) # Perform tracking on the image results = management.model.track(im0, persist=True, show=False) if results[0].boxes.id is not None: boxes = results[0].boxes.xyxy.cpu().tolist() clss = results[0].boxes.cls.cpu().tolist() management.process_data(json_data, im0, boxes, clss) # Display the processed image management.display_frames(im0) cv2.imshow("Parking Management", im0) cv2.waitKey(1) # Display each image for 1 ms cv2.destroyAllWindows() In this code: 1. We read images from a specified directory. 2. For each image, we perform the same processing as we would for a video frame. 3. The results are displayed using OpenCV's imshow. This approach should allow you to handle static images captured at regular intervals. If you have any further questions or run into issues, feel free to ask! — Reply to this email directly, view it on GitHub <#11093 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNHTQXFT7CST22E4TQSNIDZGYDHPAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMZRGQ2DK> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[glenn-jocher]

@adiyazid hi,

To improve detection accuracy and ensure that the system detects any occupied bay regardless of the vehicle type, you can modify the model to recognize a broader range of objects. Ensure your model is trained on a diverse dataset that includes various vehicle types. Additionally, you can adjust the confidence threshold and IOU settings in the model.track method to fine-tune the detection performance. For further customization, refer to the Ultralytics documentation on parking management.

[adiyazid]

Here's the availability images.

…

On Fri, Jul 19, 2024 at 10:33 AM Adi Rosli ***@***.***> wrote: Hi Paula, I've managed to use the platform for detection parking, at this moment I need to improve the accuracy Yolov8 does detect the vehicle but the results for parking availability is only based on cars, please find attached image. How do I change the code so that it improves the detection and detects an occupied bay regardless of vehicle. Please reply to me, soon, thank you. On Tue, Jun 11, 2024 at 4:22 AM Paula Derrenger ***@***.***> wrote: > Hi @adiyazid <https://github.com/adiyazid>, > > Thank you for your detailed question! It's great to hear that you've > successfully created the JSON file on Windows 10. Let's address your > concern about handling static images instead of a video stream. > > You can modify the provided code to work with a dataset of static images > by iterating through your image files instead of reading frames from a > video stream. Here's an example of how you can adapt the code to process > images from a directory: > > import cv2import osfrom ultralytics import solutions > # Path to the JSON file created with the point selection apppolygon_json_path = "bounding_boxes.json" > # Directory containing your static imagesimage_dir = "path/to/your/image/directory" > # Initialize the parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") > # Get a list of all image files in the directoryimage_files = [os.path.join(image_dir, f) for f in os.listdir(image_dir) if f.endswith(('.png', '.jpg', '.jpeg'))] > # Loop through each image filefor image_file in image_files: > # Read the image > im0 = cv2.imread(image_file) > if im0 is None: > print(f"Error reading image {image_file}") > continue > > # Extract parking regions from the JSON data > json_data = management.parking_regions_extraction(polygon_json_path) > > # Perform tracking on the image > results = management.model.track(im0, persist=True, show=False) > > if results[0].boxes.id is not None: > boxes = results[0].boxes.xyxy.cpu().tolist() > clss = results[0].boxes.cls.cpu().tolist() > management.process_data(json_data, im0, boxes, clss) > > # Display the processed image > management.display_frames(im0) > cv2.imshow("Parking Management", im0) > cv2.waitKey(1) # Display each image for 1 ms > cv2.destroyAllWindows() > > In this code: > > 1. We read images from a specified directory. > 2. For each image, we perform the same processing as we would for a > video frame. > 3. The results are displayed using OpenCV's imshow. > > This approach should allow you to handle static images captured at > regular intervals. If you have any further questions or run into issues, > feel free to ask! > > — > Reply to this email directly, view it on GitHub > <#11093 (reply in thread)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ACNHTQXFT7CST22E4TQSNIDZGYDHPAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMZRGQ2DK> > . > You are receiving this because you were mentioned.Message ID: > ***@***.*** > com> >

[pderrenger]

@adiyazid to improve the detection of occupied parking bays regardless of vehicle type, you can adjust the model to recognize various types of vehicles. Ensure your dataset includes diverse vehicle types and retrain the model if necessary. Additionally, you can modify the classes argument in the model.track method to include all relevant vehicle classes. For more detailed guidance, please refer to the Parking Management Guide.

[adiyazid]

Here's the links to the files. https://drive.google.com/file/d/12iwJxsyk6npqHeZjl4K4qGqBrq32KQrU/view?usp=sharing , https://drive.google.com/file/d/1S2W0g7JuH-yks9QX0RFXft8_iPs60FgD/view?usp=sharing , https://drive.google.com/file/d/1ZuLKX1ZvMNTkNAFSOWo3asPoLa23f6TM/view?usp=sharing

…

On Fri, Jul 19, 2024 at 10:38 AM Adi Rosli ***@***.***> wrote: Here's the availability images. On Fri, Jul 19, 2024 at 10:33 AM Adi Rosli ***@***.***> wrote: > Hi Paula, > > I've managed to use the platform for detection parking, at this moment I > need to improve the accuracy > > Yolov8 does detect the vehicle but the results for parking availability > is only based on cars, please find attached image. > > How do I change the code so that it improves the detection and detects an > occupied bay regardless of vehicle. > > Please reply to me, soon, thank you. > > On Tue, Jun 11, 2024 at 4:22 AM Paula Derrenger ***@***.***> > wrote: > >> Hi @adiyazid <https://github.com/adiyazid>, >> >> Thank you for your detailed question! It's great to hear that you've >> successfully created the JSON file on Windows 10. Let's address your >> concern about handling static images instead of a video stream. >> >> You can modify the provided code to work with a dataset of static images >> by iterating through your image files instead of reading frames from a >> video stream. Here's an example of how you can adapt the code to process >> images from a directory: >> >> import cv2import osfrom ultralytics import solutions >> # Path to the JSON file created with the point selection apppolygon_json_path = "bounding_boxes.json" >> # Directory containing your static imagesimage_dir = "path/to/your/image/directory" >> # Initialize the parking management objectmanagement = solutions.ParkingManagement(model_path="yolov8n.pt") >> # Get a list of all image files in the directoryimage_files = [os.path.join(image_dir, f) for f in os.listdir(image_dir) if f.endswith(('.png', '.jpg', '.jpeg'))] >> # Loop through each image filefor image_file in image_files: >> # Read the image >> im0 = cv2.imread(image_file) >> if im0 is None: >> print(f"Error reading image {image_file}") >> continue >> >> # Extract parking regions from the JSON data >> json_data = management.parking_regions_extraction(polygon_json_path) >> >> # Perform tracking on the image >> results = management.model.track(im0, persist=True, show=False) >> >> if results[0].boxes.id is not None: >> boxes = results[0].boxes.xyxy.cpu().tolist() >> clss = results[0].boxes.cls.cpu().tolist() >> management.process_data(json_data, im0, boxes, clss) >> >> # Display the processed image >> management.display_frames(im0) >> cv2.imshow("Parking Management", im0) >> cv2.waitKey(1) # Display each image for 1 ms >> cv2.destroyAllWindows() >> >> In this code: >> >> 1. We read images from a specified directory. >> 2. For each image, we perform the same processing as we would for a >> video frame. >> 3. The results are displayed using OpenCV's imshow. >> >> This approach should allow you to handle static images captured at >> regular intervals. If you have any further questions or run into issues, >> feel free to ask! >> >> — >> Reply to this email directly, view it on GitHub >> <#11093 (reply in thread)>, >> or unsubscribe >> <https://github.com/notifications/unsubscribe-auth/ACNHTQXFT7CST22E4TQSNIDZGYDHPAVCNFSM6AAAAABHCDMGSGVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TOMZRGQ2DK> >> . >> You are receiving this because you were mentioned.Message ID: >> ***@***.*** >> com> >> >

[pderrenger]

@adiyazid thank you for sharing the links and providing detailed context. To improve detection accuracy and ensure the system detects any occupied bay regardless of the vehicle type, you can modify the model's class filtering. Ensure your model is trained to recognize all relevant vehicle types and adjust the classes argument in the model.track method to include all these classes. For further guidance on customizing your parking management system, please refer to our comprehensive documentation at https://docs.ultralytics.com/guides/parking-management/. If you encounter any issues, make sure to check for reproducibility with the latest package versions.

[GeoJere01]

hello team im facing this issue: PS C:\Users\AJERE\Desktop\packing_management> & C:/Users/AJERE/AppData/Local/Programs/Python/Python310/python.exe c:/Users/AJERE/Desktop/packing_management/parking.py
requirements: Ultralytics requirement ['shapely>=2.0.0'] not found, attempting AutoUpdate...

[notice] A new release of pip is available: 24.0 -> 24.2
[notice] To update, run: python.exe -m pip install --upgrade pip
Requirement already satisfied: shapely>=2.0.0 in c:\python312\lib\site-packages (2.0.5)
Requirement already satisfied: numpy<3,>=1.14 in c:\python312\lib\site-packages (from shapely>=2.0.0) (2.0.1)

requirements: AutoUpdate success ✅ 44.3s, installed 1 package: ['shapely>=2.0.0']
requirements: ⚠️ Restart runtime or rerun command for updates to take effect

Traceback (most recent call last):
File "c:\Users\AJERE\Desktop\packing_management\parking.py", line 1, in
from ultralytics import solutions
File "C:\Users\AJERE\AppData\Local\Programs\Python\Python310\lib\site-packages\ultralytics\solutions_init_.py", line 6, in
from .heatmap import Heatmap
File "C:\Users\AJERE\AppData\Local\Programs\Python\Python310\lib\site-packages\ultralytics\solutions\heatmap.py", line 13, in
from shapely.geometry import LineString, Point, Polygon
ModuleNotFoundError: No module named 'shapely'
PS C:\Users\AJERE\Desktop\packing_management> & C:/Users/AJERE/AppData/Local/Programs/Python/Python310/python.exe c:/Users/AJERE/Desktop/packing_management/parking.py
requirements: Ultralytics requirement ['shapely>=2.0.0'] not found, attempting AutoUpdate...

[notice] A new release of pip is available: 24.0 -> 24.2
[notice] To update, run: python.exe -m pip install --upgrade pip
Requirement already satisfied: shapely>=2.0.0 in c:\python312\lib\site-packages (2.0.5)
Requirement already satisfied: numpy<3,>=1.14 in c:\python312\lib\site-packages (from shapely>=2.0.0) (2.0.1)
requirements: AutoUpdate success ✅ 49.9s, installed 1 package: ['shapely>=2.0.0']
requirements: ⚠️ Restart runtime or rerun command for updates to take effect

Traceback (most recent call last):
File "c:\Users\AJERE\Desktop\packing_management\parking.py", line 1, in
from ultralytics import solutions
File "C:\Users\AJERE\AppData\Local\Programs\Python\Python310\lib\site-packages\ultralytics\solutions_init_.py", line 6, in
from .heatmap import Heatmap
File "C:\Users\AJERE\AppData\Local\Programs\Python\Python310\lib\site-packages\ultralytics\solutions\heatmap.py", line 13, in
from shapely.geometry import LineString, Point, Polygon
ModuleNotFoundError: No module named 'shapely'

[glenn-jocher]

@GeoJere01 it appears that the shapely module is not being recognized despite being installed. This issue might be due to multiple Python environments. Ensure that shapely is installed in the same environment where you are running your script. You can check this by running pip show shapely in your terminal. If the issue persists, try reinstalling shapely using the command pip install shapely in the active environment. Additionally, ensure you are using the latest version of the Ultralytics package. If the problem continues, please let us know.

[GeoJere01]

how can i best mark the points? do i have to mark all parking slots or i can select a few?

[glenn-jocher]

You can mark as many or as few parking slots as you need. The more slots you mark, the more accurate the system will be in monitoring parking availability. If you only need to monitor specific areas, you can select just those slots. For detailed instructions on how to define parking spaces, please refer to the Ultralytics Parking Management Guide.

[GeoJere01]

Thank you! I tried marking a lot of slots but after running the script for the final video output all the bounding boxes are placed where i didn't mark them? Why is that? Thank you:)

[glenn-jocher]

@GeoJere01 it sounds like the bounding boxes might not be aligning correctly with the marked slots. Please ensure that the JSON file with the parking regions is correctly saved and loaded. Also, verify that the coordinates in the JSON file match the intended parking slots. If the issue persists, try updating to the latest version of the Ultralytics package and rerun the script. If you need further assistance, please refer to the Ultralytics Parking Management Guide.