Base

.gitignore

@@ -0,0 +1 @@
+.vscode/

.gitmodules

@@ -0,0 +1,7 @@
+[submodule "yolov5"]
+	path = yolov5
+	url = https://github.com/ultralytics/yolov5.git
+
+[submodule "deep_sort/deep/reid"]
+	path = deep_sort/deep/reid
+	url = https://github.com/KaiyangZhou/deep-person-reid

MOT16_eval/eval.sh

@@ -0,0 +1,109 @@
+#!/bin/bash
+
+set +e
+alias wget='curl -O'
+
+# start from clean slate
+# for i in data.zip MOT16.zip
+# do
+# 	zip -T $i
+# 	if [ $? -eq 0 ]
+# 	then
+# 		echo 'zip is ok'
+
+# 	else
+# 		echo 'zip corrupted, deleting'
+# 		rm -rf $i
+# 	fi
+# done
+
+
+# create output folder if it doesn't exist
+if [ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/inference/output ]
+then
+	mkdir -p D:/Curveball/Yolov5_DeepSort_Pytorch/inference/output
+	echo 'inference output folder created'
+fi
+
+
+
+# clone evaluation repo if it does not exist
+# if [ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval ]
+# then
+# 	echo 'Cloning official MOT16 evaluation repo'
+# 	git clone https://github.com/JonathonLuiten/TrackEval D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval
+# 	# download quick start data folder if it does not exist
+# 	if [ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data ]
+# 	then
+# 		# download data
+# 		wget -nc https://omnomnom.vision.rwth-aachen.de/data/TrackEval/data.zip -O D:/Curveball/Yolov5_DeepSort_Pytorch/data.zip
+# 		# unzip
+# 		unzip -q D:/Curveball/Yolov5_DeepSort_Pytorch/data.zip -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/
+# 		# delete zip
+# 		#rm data.zip
+# 	fi
+# fi
+
+
+# create MOT16 folder if it doesn't exist
+# if [ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16 ]
+# then
+# 	mkdir -p D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16
+# fi
+
+
+# if MOT16 data not unziped, then download, unzip and lastly remove zip MOT16 data
+# if [[ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/train ]] && [[ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/test ]]
+# then
+# 	# download data
+# 	wget -nc https://motchallenge.net/data/MOT16.zip -O D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16.zip
+# 	# unzip
+#     unzip -q MOT16.zip -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/
+# 	# delete zip
+# 	#rm MOT16.zip
+# fi
+
+
+# create folder to place tracking results for this method
+mkdir -p D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/trackers/mot_challenge/MOT16-train/ch_yolov5m_deep_sort/data/
+
+# inference on 4 MOT16 video sequences at the same time
+# suits a 4GB GRAM GPU, feel free to increase if you have more memory
+N=2
+
+# generate tracking results for each sequence
+for i in MOT16-02 MOT16-04 MOT16-05 MOT16-09 MOT16-10 MOT16-11 MOT16-13 
+do
+	(
+		# change name to inference source so that each thread write to its own .txt file
+		if [ ! -d D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/train/$i/$i ]
+		then
+			mv D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/train/$i/img1/ D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/train/$i/$i
+		fi
+		# run inference on sequence frames
+		python track.py --source D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/MOT16/train/$i/$i --save-txt --evaluate --yolo_model yolov5/weights/crowdhuman_yolov5m.pt --classes 0 --exist-ok
+	    # move generated results to evaluation repo
+	) &
+	# https://unix.stackexchange.com/questions/103920/parallelize-a-bash-for-loop
+	# allow to execute up to $N jobs in parallel
+    if [[ $(jobs -r -p | wc -l) -ge $N ]]
+	then
+        # now there are $N jobs already running, so wait here for any job
+        # to be finished so there is a place to start next one.
+        wait -n
+    fi
+done
+
+# no more jobs to be started but wait for pending jobs
+# (all need to be finished)
+wait
+echo "Inference on all MOT16 sequences DONE"
+
+echo "Moving data from experiment folder to MOT16"
+mv D:/Curveball/Yolov5_DeepSort_Pytorch/runs/track/exp/* \
+   D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/data/trackers/mot_challenge/MOT16-train/ch_yolov5m_deep_sort/data/
+
+# run the evaluation
+python D:/Curveball/Yolov5_DeepSort_Pytorch/MOT16_eval/TrackEval/scripts/run_mot_challenge.py --BENCHMARK MOT16 \
+ --TRACKERS_TO_EVAL ch_yolov5m_deep_sort --SPLIT_TO_EVAL train --METRICS CLEAR Identity \
+ --USE_PARALLEL False --NUM_PARALLEL_CORES 4

README.md

@@ -0,0 +1,63 @@
+# Pedestrian Speed Estimation using yolov5, deepsort and homography matrix
+
+## Demo gif
+
+<img src="MOT16_eval/demo.gif" width="600"/>
+
+## Before you run the tracker
+
+1. Clone the repository
+
+2. Install dependencies
+
+`pip install -r requirements.txt`
+
+## Download weights 
+
+Download [weights](https://drive.google.com/file/d/1gglIwqxaH2iTvy6lZlXuAcMpd_U0GCUb/view?usp=sharing) and place it in yolov5/weights folder
+
+## Quick run
+Add a video to data/videos folder and run the following command.
+
+```bash
+python track.py --source data/videos/test.mp4 --yolo_model yolov5/weights/crowdhuman_yolov5m.pt --classes 0 --show-vid --save-vid
+```
+
+Results are saved to folder `track/expN`
+
+## Tracking sources
+
+Tracking can be run on most video formats
+
+```bash
+$ python track.py --source 0  # webcam
+                           img.jpg  # image
+                           vid.mp4  # video
+                           path/  # directory
+                           path/*.jpg  # glob
+                           'https://youtu.be/Zgi9g1ksQHc'  # YouTube
+                           'rtsp://example.com/media.mp4'  # RTSP, RTMP, HTTP stream
+```
+
+### DeepSort
+
+Choose a ReID model based on your needs from this ReID [model zoo](https://kaiyangzhou.github.io/deep-person-reid/MODEL_ZOO)
+
+```bash
+
+
+$ python track.py --source 0 --deep_sort_model osnet_x1_0
+                                               nasnsetmobile
+                                               resnext101_32x8d
+                                               ...
+```
+## Reference
+
+```latex
+@misc{yolov5deepsort2020,
+    title={Real-time multi-object tracker using YOLOv5 and deep sort},
+    author={Mikel Broström},
+    howpublished = {\url{https://github.com/mikel-brostrom/Yolov5_DeepSort_Pytorch}},
+    year={2020}
+}
+```

deep_sort/.gitignore

@@ -0,0 +1,13 @@
+# Folders
+__pycache__/
+build/
+*.egg-info
+
+
+# Files
+*.weights
+*.t7
+*.mp4
+*.avi
+*.so
+*.txt

deep_sort/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2020 Ziqiang
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

deep_sort/README.md

@@ -0,0 +1,137 @@
+# Deep Sort with PyTorch
+
+![](demo/demo.gif)
+
+## Update(1-1-2020)
+Changes
+- fix bugs
+- refactor code
+- accerate detection by adding nms on gpu
+
+## Latest Update(07-22)
+Changes
+- bug fix (Thanks @JieChen91 and @yingsen1 for bug reporting).  
+- using batch for feature extracting for each frame, which lead to a small speed up.  
+- code improvement.
+
+Futher improvement direction  
+- Train detector on specific dataset rather than the official one.
+- Retrain REID model on pedestrain dataset for better performance.
+- Replace YOLOv3 detector with advanced ones.
+
+**Any contributions to this repository is welcome!**
+
+
+## Introduction
+This is an implement of MOT tracking algorithm deep sort. Deep sort is basicly the same with sort but added a CNN model to extract features in image of human part bounded by a detector. This CNN model is indeed a RE-ID model and the detector used in [PAPER](https://arxiv.org/abs/1703.07402) is FasterRCNN , and the original source code is [HERE](https://github.com/nwojke/deep_sort).  
+However in original code, the CNN model is implemented with tensorflow, which I'm not familier with. SO I re-implemented the CNN feature extraction model with PyTorch, and changed the CNN model a little bit. Also, I use **YOLOv3** to generate bboxes instead of FasterRCNN.
+
+## Dependencies
+- python 3 (python2 not sure)
+- numpy
+- scipy
+- opencv-python
+- sklearn
+- torch >= 0.4
+- torchvision >= 0.1
+- pillow
+- vizer
+- edict
+
+## Quick Start
+0. Check all dependencies installed
+```bash
+pip install -r requirements.txt
+```
+for user in china, you can specify pypi source to accelerate install like:
+```bash
+pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple
+```
+
+1. Clone this repository
+```
+git clone git@github.com:ZQPei/deep_sort_pytorch.git
+```
+
+2. Download YOLOv3 parameters
+```
+cd detector/YOLOv3/weight/
+wget https://pjreddie.com/media/files/yolov3.weights
+wget https://pjreddie.com/media/files/yolov3-tiny.weights
+cd ../../../
+```
+
+3. Download deepsort parameters ckpt.t7
+```
+cd deep_sort/deep/checkpoint
+# download ckpt.t7 from
+https://drive.google.com/drive/folders/1xhG0kRH1EX5B9_Iz8gQJb7UNnn_riXi6 to this folder
+cd ../../../
+```  
+
+4. Compile nms module
+```bash
+cd detector/YOLOv3/nms
+sh build.sh
+cd ../../..
+```
+
+Notice:
+If compiling failed, the simplist way is to **Upgrade your pytorch >= 1.1 and torchvision >= 0.3" and you can avoid the troublesome compiling problems which are most likely caused by either `gcc version too low` or `libraries missing`.
+
+5. Run demo
+```
+usage: python yolov3_deepsort.py VIDEO_PATH
+                                [--help]
+                                [--frame_interval FRAME_INTERVAL]
+                                [--config_detection CONFIG_DETECTION]
+                                [--config_deepsort CONFIG_DEEPSORT]
+                                [--display]
+                                [--display_width DISPLAY_WIDTH]
+                                [--display_height DISPLAY_HEIGHT]
+                                [--save_path SAVE_PATH]          
+                                [--cpu]          
+
+# yolov3 + deepsort
+python yolov3_deepsort.py [VIDEO_PATH]
+
+# yolov3_tiny + deepsort
+python yolov3_deepsort.py [VIDEO_PATH] --config_detection ./configs/yolov3_tiny.yaml
+
+# yolov3 + deepsort on webcam
+python3 yolov3_deepsort.py /dev/video0 --camera 0
+
+# yolov3_tiny + deepsort on webcam
+python3 yolov3_deepsort.py /dev/video0 --config_detection ./configs/yolov3_tiny.yaml --camera 0
+```
+Use `--display` to enable display.  
+Results will be saved to `./output/results.avi` and `./output/results.txt`.
+
+All files above can also be accessed from BaiduDisk!  
+linker：[BaiduDisk](https://pan.baidu.com/s/1YJ1iPpdFTlUyLFoonYvozg)
+passwd：fbuw
+
+## Training the RE-ID model
+The original model used in paper is in original_model.py, and its parameter here [original_ckpt.t7](https://drive.google.com/drive/folders/1xhG0kRH1EX5B9_Iz8gQJb7UNnn_riXi6).  
+
+To train the model, first you need download [Market1501](http://www.liangzheng.com.cn/Project/project_reid.html) dataset or [Mars](http://www.liangzheng.com.cn/Project/project_mars.html) dataset.  
+
+Then you can try [train.py](deep_sort/deep/train.py) to train your own parameter and evaluate it using [test.py](deep_sort/deep/test.py) and [evaluate.py](deep_sort/deep/evalute.py).
+![train.jpg](deep_sort/deep/train.jpg)
+
+## Demo videos and images
+[demo.avi](https://drive.google.com/drive/folders/1xhG0kRH1EX5B9_Iz8gQJb7UNnn_riXi6)
+[demo2.avi](https://drive.google.com/drive/folders/1xhG0kRH1EX5B9_Iz8gQJb7UNnn_riXi6)
+
+![1.jpg](demo/1.jpg)
+![2.jpg](demo/2.jpg)
+
+
+## References
+- paper: [Simple Online and Realtime Tracking with a Deep Association Metric](https://arxiv.org/abs/1703.07402)
+
+- code: [nwojke/deep_sort](https://github.com/nwojke/deep_sort)
+
+- paper: [YOLOv3](https://pjreddie.com/media/files/papers/YOLOv3.pdf)
+
+- code: [Joseph Redmon/yolov3](https://pjreddie.com/darknet/yolo/)

deep_sort/__init__.py

@@ -0,0 +1,11 @@
+from .deep_sort import DeepSort
+
+
+__all__ = ['DeepSort', 'build_tracker']
+
+
+def build_tracker(cfg, use_cuda):
+    return DeepSort(cfg.DEEPSORT.REID_CKPT, 
+                max_dist=cfg.DEEPSORT.MAX_DIST, min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE, 
+                nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP, max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE, 
+                max_age=cfg.DEEPSORT.MAX_AGE, n_init=cfg.DEEPSORT.N_INIT, nn_budget=cfg.DEEPSORT.NN_BUDGET, use_cuda=use_cuda)

deep_sort/configs/deep_sort.yaml

@@ -0,0 +1,8 @@
+DEEPSORT:
+  MODEL_TYPE: "osnet_x1_0"
+  MAX_DIST: 0.2 # The matching threshold. Samples with larger distance are considered an invalid match
+  MAX_IOU_DISTANCE: 0.7 # Gating threshold. Associations with cost larger than this value are disregarded.
+  MAX_AGE: 30 # Maximum number of missed misses before a track is deleted
+  N_INIT: 3 # Number of frames that a track remains in initialization phase
+  NN_BUDGET: 100 # Maximum size of the appearance descriptors gallery
+