How to Read Frames from Videocapture from Secondary Webcam with Opencv

Unable to read frames from VideoCapture from secondary webcam with OpenCV

After some time I've found out that it is always only the first call of read that fails and skipping the first frame started to work fine although the true reason of this behavior remained unknown.

Later James Barnett (see comments above) has pointed out that the reason might be that it takes a while till the camera gets ready for capturing and my current solution looks the following way (C++11's sleep):

#include <chrono>
#include <thread>
...

VideoCapture cap(1);

// give camera some extra time to get ready:
std::this_thread::sleep_for(std::chrono::milliseconds(200));

if (!cap.isOpened()) {
     std::cout << "Unable to read stream from specified device." << std::endl;
     return;
}

while (true)
{
    // retrieve the frame:
    Mat frame;
    if (!cap.read(frame)) {
        std::cout << "Unable to retrieve frame from video stream." << std::endl;
        continue;
    }

    // display it:
    imshow("LiveStream", frame);

    // stop if Esc has been pressed:
    if (waitKey(1) == 27) {
        break;
    }
}

cap.release();

Hopefully some future visitors will find it helpful :)

Reading every nth frame from VideoCapture in OpenCV

I'm afraid there's not much you can do and it's not just a shortcoming of OpenCV. You see, modern video codecs, are, generally, complex beasts. To get a higher compression rate the encoding of a frame is often dependent on previous and sometimes even successive frames.

So, most of the time you have to decode frames before the desired one even if you don't need them.

There are rather non-trivial tricks to specifically encode a video file, so that it would be cheap to get every Nth frame, but it's not feasible in general case.

That said, you can try the seeking functionality OpenCV provides (see OpenCV Seek Function/Rewind). It may (as well as may not) work faster depending on the circumstances. However, personally, I wouldn't bet on it.

OpenCV/Python: read specific frame using VideoCapture

Thank you GPPK.

The video parameters should be given as integers. Each flag has its own value. See here for the code.

The correct solution is:

import numpy as np
import cv2

#Get video name from user
#Ginen video name must be in quotes, e.g. "pirkagia.avi" or "plaque.avi"
video_name = input("Please give the video name including its extension. E.g. \"pirkagia.avi\":\n")

#Open the video file
cap = cv2.VideoCapture(video_name)

#Set frame_no in range 0.0-1.0
#In this example we have a video of 30 seconds having 25 frames per seconds, thus we have 750 frames.
#The examined frame must get a value from 0 to 749.
#For more info about the video flags see here: https://stackoverflow.com/questions/11420748/setting-camera-parameters-in-opencv-python
#Here we select the last frame as frame sequence=749. In case you want to select other frame change value 749.
#BE CAREFUL! Each video has different time length and frame rate. 
#So make sure that you have the right parameters for the right video!
time_length = 30.0
fps=25
frame_seq = 749
frame_no = (frame_seq /(time_length*fps))

#The first argument of cap.set(), number 2 defines that parameter for setting the frame selection.
#Number 2 defines flag CV_CAP_PROP_POS_FRAMES which is a 0-based index of the frame to be decoded/captured next.
#The second argument defines the frame number in range 0.0-1.0
cap.set(2,frame_no);

#Read the next frame from the video. If you set frame 749 above then the code will return the last frame.
ret, frame = cap.read()

#Set grayscale colorspace for the frame. 
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

#Cut the video extension to have the name of the video
my_video_name = video_name.split(".")[0]

#Display the resulting frame
cv2.imshow(my_video_name+' frame '+ str(frame_seq),gray)

#Set waitKey 
cv2.waitKey()

#Store this frame to an image
cv2.imwrite(my_video_name+'_frame_'+str(frame_seq)+'.jpg',gray)

# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()

Capturing video from two cameras in OpenCV at once

Yes you're definitely limited by the USB bandwidth. Attempting to read from both devices at full-rez you probably got error:

libv4l2: error turning on stream: No space left on device
VIDIOC_STREAMON: No space left on device
Traceback (most recent call last):
  File "p.py", line 7, in <module>
    assert ret1 # fails?!
AssertionError

And then when you reduce the res to 160x120:

import cv2
cap0 = cv2.VideoCapture(0)
cap0.set(3,160)
cap0.set(4,120)
cap1 = cv2.VideoCapture(1)
cap1.set(3,160)
cap1.set(4,120)
ret0, frame0 = cap0.read()
assert ret0 # succeeds
ret1, frame1 = cap1.read()
assert ret1 # fails?!

now it seems to work! I bet you have both cams connected on the same USB card. You can run lsusb command to make sure, and it should indicate something like:

Bus 001 Device 006: ID 046d:081b Logitech, Inc. Webcam C310
Bus 001 Device 004: ID 0409:005a NEC Corp. HighSpeed Hub
Bus 001 Device 007: ID 046d:0990 Logitech, Inc. QuickCam Pro 9000
Bus 001 Device 005: ID 0409:005a NEC Corp. HighSpeed Hub
Bus 001 Device 003: ID 0409:005a NEC Corp. HighSpeed Hub
Bus 001 Device 002: ID 1058:0401 Western Digital Technologies, Inc. 
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub

(Note both cameras on same bus.) If possible, you can add another USB card to your machine to gain more bandwidth. I've done this before in order to run multiple cams at full resolution on a single machine. Albeit that was a tower workstation with available motherboard slots, and unfortunately you may not have that option on a MacBook laptop.

How to get the latest frame from capture device (camera) in opencv

I think the solution mentioned in the question, namely having a separate thread that clears the buffer, is the easiest non-brittle solution for this. Here reasonably nice (I think) code for this:

import cv2, queue, threading, time

# bufferless VideoCapture
class VideoCapture:

  def __init__(self, name):
    self.cap = cv2.VideoCapture(name)
    self.q = queue.Queue()
    t = threading.Thread(target=self._reader)
    t.daemon = True
    t.start()

  # read frames as soon as they are available, keeping only most recent one
  def _reader(self):
    while True:
      ret, frame = self.cap.read()
      if not ret:
        break
      if not self.q.empty():
        try:
          self.q.get_nowait()   # discard previous (unprocessed) frame
        except queue.Empty:
          pass
      self.q.put(frame)

  def read(self):
    return self.q.get()

cap = VideoCapture(0)
while True:
  time.sleep(.5)   # simulate time between events
  frame = cap.read()
  cv2.imshow("frame", frame)
  if chr(cv2.waitKey(1)&255) == 'q':
    break

The frame reader thread is encapsulated inside the custom VideoCapture class, and communication with the main thread is via a queue.

I posted very similar code for a node.js question, where a JavaScript solution would have been better. My comments on another answer to that question give details why a non-brittle solution without separate thread seems difficult.

An alternative solution that is easier but supported only for some OpenCV backends is using CAP_PROP_BUFFERSIZE. The 2.4 docs state it is "only supported by DC1394 [Firewire] v 2.x backend currently." For Linux backend V4L, according to a comment in the 3.4.5 code, support was added on 9 Mar 2018, but I got VIDEOIO ERROR: V4L: Property <unknown property string>(38) not supported by device for exactly this backend. It may be worth a try first; the code is as easy as this:

cap.set(cv2.CAP_PROP_BUFFERSIZE, 0)

---

Related Topics