show FPS yolov5

[HuumbleBee]

Hello i want to show fps yolov5 object detection on cv2, i have search how to show it, but i still not success to do it. can anyone can direct me where can i put fps computing program so that if i running detect.py fps can appear in cv2? thank you

[glenn-jocher]

@HuumbleBee just divide 1000 ms by any of these speeds to get FPS:

[glenn-jocher]

just divide 1000 ms by any of these speeds to get FPS:

[hoseinhamzei]

Speed: 0.7ms pre-process, 35.6ms inference, 2.0ms NMS per image at shape (1, 3, 640, 640)
thanks but sorry i don't understand, here are three values (speed, inference, NMS per image) which one should i divide 1000 by to get average FPS when running the detection on a video?

[glenn-jocher]

whichever you want based on your definition of FPS

[phuctrang]

in my project, i get 1000/(pre-process+inference+NMS)

[zehranrgi]

For example,
the time taken for pre-processing is 0.8ms, the time taken for inference is 13.2ms, and the time taken for NMS is 3.4ms. Therefore, the total time taken to process one image is 17.4ms.

To calculate the FPS, you need to divide the number of milliseconds in one second (1000) by the time taken to process one image in milliseconds:

FPS = 1000 / 17.4 = 57.47

So, the FPS for this YOLOv5 model on this particular image is approximately 57.47.

[chunxiao365]

Hello i want to show fps yolov5 object detection on cv2, i have search how to show it, but i still not success to do it. can anyone can direct me where can i put fps computing program so that if i running detect.py fps can appear in cv2? thank you

have you solved your question? I also want to know how

[michaelkedev]

You can use cv2.putText()

[chunxiao365]

my code is :

`#!/usr/bin/python3
import cv2
import torch
from PIL import Image
import time

cap = cv2.VideoCapture(0)
print("capture device is open: " + str(cap.isOpened()))
fps_start_time = time.time()
fps=0

while True:

# img = infence(frame)
# caculate fps
success, frame = cap.read()
fps_end_time = time.time()
fps_diff_time = fps_end_time - fps_start_time
fps = 1 / fps_diff_time
fps_start_time = fps_end_time
fps_text="FPS:{:.2f}".format(fps)
cv2.putText(frame, fps_text, (5, 30), cv2.FONT_HERSHEY_COMPLEX, 1, (0, 255, 255), 1)
cv2.imshow("webcam",frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
    break

cap.release()

`

[AbrarKhan009]

Hello everyone, i am sharing a youtube video where he guide you how to show FPS on real time running video on YOLOv7 maybe it will help you and you can modify the code for yolov5 also after that.

https://www.youtube.com/watch?v=_CkXDjmT8dc&ab_channel=TheCodingBug

[supriamir]

Hi @glenn-jocher,

I have calculate the FPS by 1000/speeds , where speed sum of (pre-processing, inference time, post-processing).
'Im encountering confusion with my FPS calculation results. Despite having a model with fewer parameters, it's exhibiting a higher FPS. What could be the reason for this?