【python】video2image

import struct
import json

import cv2 
import argparse
import os

parser = argparse.ArgumentParser()
parser.add_argument('--source_video_mp4', type=str, default='/home/momenta/Downloads/dvr/202107081536.mp4')
parser.add_argument('--source_video_h264', type=str, default='/home/momenta/Downloads/dvr/tmp.h264')
parser.add_argument('--output_dir', type=str, default='output')

opt = parser.parse_args()

videoCapture = cv2.VideoCapture()
videoCapture.open(opt.source_video_mp4)


class Frame:
    def __init__(self, frame_type, start_index):
        self.frame_type = frame_type
        self.data = []


class VideoInfo:
    def __init__(self):
        self.timestamps = []


def parse_frame(frame, info):
    if frame.data[0] == 0x06 and frame.data[1] == 0x5 and frame.data[2] == 0x20:
        # for b in frame.data:
        #     logger.info(hex(b), end='')
        # logger.info()
        uuid, timestamp = struct.unpack('16sQ', bytes(frame.data[3:27]))
        print(frame.data[3:27])
        info.timestamps.append(timestamp)
    
if __name__ == '__main__':
    info = VideoInfo()
    with open(opt.source_video_h264, 'rb') as reader:
        key = 0
        count = 0
        key_found = False
        frame = None
        data = reader.read()

        print(len(data))

        for i in range(0, len(data) - 1):
            b = data[i]
            if key_found:
                key_found = False
                frame_type = b & 0x1F
                frame = Frame(frame_type, i)
                frame.data.append(b)
            else:
                if frame:
                    frame.data.append(b)
                key = (key << 8) | b
                key &= 0xFFFFFFFF
                if key & 0xFFFFFF == 0x000001:
                    count += 1
                    key_found = True
                    if frame:
                        if key == 0x000000001:
                            frame.data = frame.data[:-4]
                        else:
                            frame.data = frame.data[:-3]
                        parse_frame(frame, info)
    print(info.timestamps)
    print(len(info.timestamps))
    ind = [i for i in range(len(info.timestamps))]
    to_save = dict(zip(ind, info.timestamps))

    fps = videoCapture.get(cv2.CAP_PROP_FPS)
    frames = videoCapture.get(cv2.CAP_PROP_FRAME_COUNT)
    print("fps=",fps,"frames=",frames)
    print(int(frames))
    f1 = open("{}/timestamp.txt".format(opt.output_dir),'w')
    for i in range(int(frames)):
        ret,frame = videoCapture.read()
        if ret is True:
            timestamps = videoCapture.get(cv2.CAP_PROP_POS_MSEC)
            cv2.imwrite("{}/{}.jpg".format(opt.output_dir, to_save[i]), frame)
            f1.writelines("{}".format(timestamps))
    f1.close()