Split scenes with ffmpeg

split_scenes.py

import subprocess
import sys
from pathlib import Path


def get_segment_times(video_file):
    detect_scene_command = [
        "ffmpeg",
        "-i",
        str(video_file),
        "-filter_complex",
        "select='gt(scene,0.2)',metadata=print:file=-",
        "-f",
        "null",
        "-",
    ]

    def parse_frame_info(line):
        return dict([x.split(':') for x in line.split()])

    def parse_frame_time(line):
        x = parse_frame_info(line)
        return float(x['pts_time'])

    output = subprocess.check_output(detect_scene_command)
    lines = output.decode().split('\n')
    lines = [x for x in lines if x.startswith("frame")]
    segment_times = [parse_frame_time(x) for x in lines]
    return segment_times


def split_video(video_file, segment_times, target_files):
    def segment_times_to_string(times):
        return ','.join(str(x) for x in times)

    segment_times_string = segment_times_to_string(segment_times)

    segment_command = [
        "ffmpeg",
        "-i",
        str(video_file),
        "-force_key_frames:v",
        segment_times_string,
        "-f",
        "segment",
        "-segment_times",
        segment_times_string,
        "-reset_timestamps",
        "1",
        "-c:v",
        "libx264",
        "-c:a",
        "copy",
        "-crf",
        "10",
        str(target_files),
    ]

    subprocess.call(segment_command)


input_dir = Path(sys.argv[1])
output_dir = Path(sys.argv[2])

for video in input_dir.glob('*.mp4'):
    target_name = output_dir / f"{Path(video).stem}_%03d.mp4"
    segment_times = get_segment_times(video)
    split_video(video, segment_times, target_name)