文件目录中多个视频的帧提取

Question

虽然存在多种从视频中提取帧的方法，但我想创建一个loop 用于对目录中存在的多个视频进行帧提取。

要包含的一个主要关键功能是以串行方式（即 1-60 然后 61-120 及更多）对帧进行编号。 一种方法是一次提取一个视频并重复所有视频，但这很长。

尝试过的代码：（从某个地方得到它并尝试根据我的目录进行更改）

i = 1  # Counter of first video
input_filenames='Dataset/videos'

for input_filename in intput_filenames:
    cap = cv2.VideoCapture(input_filename)

   
    while True:
        ret, frame = cap.read()  # Read frame from first video

        if ret:
            cv2.imwrite(str(i) + '.jpg', frame)  
            cv2.imshow('frame', frame)  # Displaying frame for testing
            i += 1 # Advance file counter
        else:
            break # Break the interal loop when res status is False.

        cv2.waitKey(100) #Wait 100msec 

    cap.release() 

Answer 1

经过大量的搜索和尝试，终于达到了迭代目录下的视频文件，连续提取帧的目的。

代码：

def frames_multiple_videos(filenames):
    i = 1  # Counter of first video

    for j in filenames:
        cap = cv2.VideoCapture(j)

        while True:
            ret, frame = cap.read()  # Read frame from first video
        
            if ret!=True:
                cv2.imwrite(filenames +'frame' + str(i) + '.jpg', frame)  # Write frame to JPEG file (frame1.jpg, frame2.jpg, ...)
                i += 1 # next file counter
            else:
                break

            cv2.waitKey(100) #(for debugging)

        cap.release() #Release must be inside the outer loop
        
all_frame=frames_multiple_videos('*your_directory*')

Answer 2

如果我猜对了，您希望以特定帧数提取目录中的每个视频，对吧？ 你可以试试这个代码！

for line in f.readlines(): #reading each video in directory, in this case I save in txt and read each line
    frame_reader = imageio.get_reader(link_video, 'ffmpeg', mode='I')
    start_idx, end_idx = 10, 70
    save_mat = np.full((output_size, output_size, 3, 1), 0) #please define the size of frame
    for frame_num, frame in enumerate(frame_reader):
        if frame_num >= end_idx:
           break
        if frame_num <= start_idx:
           continue
        save_mat = np.concatenate((save_mat, np.expand_dims(frame, axis=3)), axis=3)