Recherche avancée

Sur d’autres sites (10195)

• how to make cv2.videoCapture.read() faster ?

  9 novembre 2022, par Yu-Long Tsai

  My question :

  



  I was working on my computer vision project. I use opencv(4.1.2) and python to implement it.

  



  I need a faster way to pass the reading frame into image processing on my Computer(Ubuntu 18.04 8 cores i7 3.00GHz Memory 32GB). the cv2.VideoCapture.read() read frame (frame size : 720x1280) will take about 120 140ms. which is too slow. my processing module take about 40ms per run. And we desire 25 30 FPS.

  



  here is my demo code so far :

  



  import cv2
from collections import deque
from time import sleep, time
import threading


class camCapture:
    def __init__(self, camID, buffer_size):
        self.Frame = deque(maxlen=buffer_size)
        self.status = False
        self.isstop = False
        self.capture = cv2.VideoCapture(camID)


    def start(self):
        print('camera started!')
        t1 = threading.Thread(target=self.queryframe, daemon=True, args=())
        t1.start()

    def stop(self):
        self.isstop = True
        print('camera stopped!')

    def getframe(self):
        print('current buffers : ', len(self.Frame))
        return self.Frame.popleft()

    def queryframe(self):
        while (not self.isstop):
            start = time()
            self.status, tmp = self.capture.read()
            print('read frame processed : ', (time() - start) *1000, 'ms')
            self.Frame.append(tmp)

        self.capture.release()

cam = camCapture(camID=0, buffer_size=50)
W, H = 1280, 720
cam.capture.set(cv2.CAP_PROP_FRAME_WIDTH, W)
cam.capture.set(cv2.CAP_PROP_FRAME_HEIGHT, H)


# start the reading frame thread
cam.start()

# filling frames
sleep(5)

while True:
  frame = cam.getframe() # numpy array shape (720, 1280, 3)

  cv2.imshow('video',frame)
  sleep( 40 / 1000) # mimic the processing time

  if cv2.waitKey(1) == 27:
        cv2.destroyAllWindows()
        cam.stop()
        break



  



  What I tried :

  



  


  1. multiThread - one thread just reading the frame, the other do the image processing things.
It's NOT what I want. because I could set a buffer deque saving 50 frames for example. but the frame-reading thread worked with the speed frame/130ms. my image processing thread worked with the speed frame/40ms. then the deque just running out. so I've been tried the solution. but not what I need.

  


  2. this topic is the discussion I found out which is most closest to my question. but unfortunately, I tried the accepted solutions (both of two below the discussion).

  


  



  One of the solution (6 six thumbs up) point out that he could reading and saving 100 frames at 1 sec intervals on his mac. why my machine cannot handle the frame reading work ? Do I missing something ? my installation used conda and pip conda install -c conda-forge opencv, pip install opencv-python(yes, I tried both.)

  



  The other of the solution(1 thumb up) using ffmpeg solution. but it seem's work with video file but not camera device ?

  



  

  3. adjust c2.waitKey() : 
the parameter just controls the frequency when video display. not a solution.

  


  



  Then, I know I just need some keywords to follow.

  



  code above is my demo code so far, I want some method or guide to make me videoCapture.read() faster. maybe a way to use multithread inside videoCapture object or other camera reading module.

  



  Any suggestions ?

  


• Loss packets from RTP streaming decoded with ffmpeg

  29 avril 2020, par anaVC94

  Hope someone could help me. I am trying to apply a neural network (YOLOv2) to perform object detection to a RTP stream which I have simulated in local using VLC, with the "using RTP over TCP" mark as true.
The stream is 4K, 30fps, 15Mb/s. I am using OpenCV C API I/O module to read frames from the stream.

  



  I am using the common code. I open the RTP as follows :
cap = cvCaptureFromFile(url);

  



  And then I perform capture in one thread like :
IplImage* src = cvQueryFrame(cap);

  



  and, on another thread, the detection part.

  



  I know OpenCV uses ffmpeg to capture video. I am using ffmpeg 3.3.2. My problem is that when I receive the stream, a lot of artifacts appear. The output I get is :

  



  top block unavailable for requested intra mode -1
[h264 @ 0000016ae36f0a40] error while decoding MB 40 0, bytestream 81024
[h264 @ 0000016ae32f3860] top block unavailable for requested intra mode
[h264 @ 0000016ae32f3860] error while decoding MB 48 0, bytestream 102909
[h264 @ 0000016ae35e9e60] Reference 3 >= 3
[h264 @ 0000016ae35e9e60] error while decoding MB 79 0, bytestream 27231
[h264 @ 0000016a7033eb40] mmco: unref short failure
[h264 @ 0000016ae473ee00] Reference 3 >= 3


  



  over and over again, and there are too many packet losses that I can't see anything when showing the received frames. However, it doesn't happen to me when I stream over RTP other lower quality videos such as HD in 30fps or like that. It is also true that the 4K has a lot of movement (it is a Moto GP Race).

  



  I have tried :
- Reduce the fps of the streaming.
- Reduce the bitrate 
- ffplay either doesn't show correctly the input frames, but VLC does (don't know why).
- Force TCP transmission.
- Force input fps doing cvSetCaptureProperty(cap, CV_CAP_PROP_FPS, 1);

  



  Why is this happening and how can I improve the packet losses ? Is there any way or something else I can try ?

  



  Thanks a lot

  


• MovieWriter FFwriter unavailable. Trying to use pillow instead

  17 février 2020, par Life is Good

  I have created a bar chart race in Matplotlib and now I am trying to save it into a gif file. I have imported the relevant library :

  import matplotlib.animation as animation
  
  plt.rcParams['animation.ffmpeg_path'] = '‪C:\\Program Files\\FFmpeg\\bin\\ffmpeg.exe'
  
  FFwriter = animation.FFMpegWriter()

  Here is the code I used to create my animation :

  fig, ax = plt.subplots(figsize=(15, 8))
  
  animator = animation.FuncAnimation(fig, draw_barchart, frames=range(1999, 2015))
  
  HTML(animator.to_jshtml()) 

  However, when I searched for the writers, there are only two of them :

  animation.writers.list()

  

  I had already installed FFmpeg following a tutorial on Wikihow, and I am also able to run FFmpeg from the command line, so I don’t understand why it is not showing up. When I try to save my animation into a gif, I get this error message :
  

  Would anybody be familiar with this error message, please ? Thank you very much.