Writing and Reading video without saving in opencv-CodePudding

I want to read a video after I write it with cv2.VideoWriter without saving the video.

For example:

video =  cv2.VideoWriter('using.mp4', cv2.VideoWriter_fourcc(*'MJPG'), 10, size)

Now, after writing this cv2.VideoWriter object, is it possible to read it likevideo.read(), but since read() is a function of cv2.VideoCapture and it will throw an error

Exception has occurred: AttributeError
'cv2.VideoWriter' object has no attribute 'read'

So, is there possible way of reading the cv2.VideoWriter?

CodePudding user response：

An alternative to reading frames from the video writer, is to save the frames in a list instead of saving each frame in the the loop. when you finished, you can write them outside the loop and have the save affect as video.read()

video =  cv2.VideoWriter('using.mp4', cv2.VideoWriter_fourcc(*'MJPG'), 10, size)
for frame in frames:
    writer.write(frame)
for frame in frames:
    # do other stuff here

detailed example (Notice i changed the fourcc - your example didnt work for me)

import cv2


def cam_test(port: int = 0) -> None:
    frames = []
    cap = cv2.VideoCapture(port)
    if not cap.isOpened():  # Check if the web cam is opened correctly
        print("failed to open cam")
    else:
        print('cam opened on port {}'.format(port))

        for i in range(10 ** 10):
            success, cv_frame = cap.read()
            if not success:
                print('failed to capture frame on iter {}'.format(i))
                break
            frames.append(cv_frame)
            cv2.imshow('Input', cv_frame)
            k = cv2.waitKey(1)
            if k == ord('q'):
                break

        cap.release()
        cv2.destroyAllWindows()

    # Now you have the frames at hand

    if len(frames) > 0:
        # if you want to write them
        size = (frames[0].shape[1], frames[0].shape[0])
        video = cv2.VideoWriter(
            filename='using.mp4',
            fourcc=cv2.VideoWriter_fourcc(c1='m', c2='p', c3='4', c4='v'),
            fps=10,
            frameSize=size
        )
        for frame in frames:
            video.write(frame)

        # and to answer your question, you wanted to do video.read() which would have gave you frame by frame
        for frame in frames:
            pass  # each iteration is like video.read() if video.read() was possible

    return


if __name__ == '__main__':
    cam_test()