when called split_video_ffmpeg throw an error

[pengyouak]

Some videos will throw this error halfway through the process

bool' object has no attribute 'shape'

[Breakthrough]

Can you provide more context? A backtrace, etc? What OS, Python, and package versions you're using? Have you tried updating OpenCV or using a different backend?

Run scenedetect version and report that. Can you also share more about your workflow?

[pengyouak]

[PySceneDetect] PySceneDetect 0.6.4

System Info

OS Windows-10-10.0.22621-SP0
Python 3.8.6

Packages

av Not Installed
click 8.1.7
cv2 4.10.0
moviepy 1.0.3
numpy 1.24.3
platformdirs 4.3.6
scenedetect 0.6.4
tqdm 4.66.5

Tools

ffmpeg 4.2.2
mkvmerge Not Installed

===================================================================================
Output

 Detected: 29 | Progress: 100%|| 2181/2181 [00:30<00:00, 70.60frames/s] Detected: 41 | Progress: 100%|| 2457/2457 [00:36<00:00, 67.27frames/s] 100%|| 90/90 [00:11<00:00, 7.58images/s] Detected: 24 | Progress: 100%|| 3603/3603 [00:53<00:00, 67.30frames/s] 7%|██████████████▌ | 5/75 [00:00<00:11, 6.27images/s]'bool' object has no attribute 'shape'████████████████████████████████| 116/126 [00:16<00:01, 8.30images/s] 97%|▏ | 73/75 [00:10<00:00, 7.64images/s]'bool' object has no attribute 'shape'████████████████████████████████████▉ | 696/2181 [00:20<00:44, 33.42frame/s] 100%|| 2181/2181 [00:33<00:00, 64.84frame/s]
100%|| 2181/2181 [00:33<00:00, 110.37frame/s]

====================================================================================
Code

def split_video_into_scenes(video_path, threshold, with_html=None):
video = open_video(video_path) video_name = os.path.basename(video_path)
video_name_without_ext = os.path.splitext(video_name)[0]
video_folder = os.path.dirname(video_path)
output_folder = os.path.join(video_folder, video_name_without_ext)
if not os.path.exists(output_folder):
os.makedirs(output_folder)
html_folder = os.path.join(output_folder, 'html')
if not os.path.exists(html_folder):
os.makedirs(html_folder)
scene_manager = SceneManager()
scene_manager.add_detector(
ContentDetector(threshold=threshold))
scene_manager.detect_scenes(video, show_progress=True)
scene_list = scene_manager.get_scene_list()
if with_html:
image_list = sm.save_images(scene_list, video, output_dir=html_folder, show_progress=True, scale=0.1)
sm.write_scene_list_html(os.path.join(html_folder, 'result.html'), scene_list, image_filenames=image_list)
split_video_ffmpeg(video_path, scene_list, output_folder, show_progress=True)
if __name__ == '__main__':
video_dir = input('please input a directory\n')
threshold = 30.0
with_html = True
if video_dir.lower().endswith('.mp4'):
split_video_into_scenes(video_dir, threshold, with_html)
else:
video_files = [os.path.join(video_dir, f) for f in os.listdir(video_dir) if f.lower().endswith('.mp4')]
with concurrent.futures.ThreadPoolExecutor(max_workers=4) as executor:
futures = [executor.submit(split_video_into_scenes, video_path, threshold, with_html) for video_path in video_files]
for future in concurrent.futures.as_completed(futures):
try:
future.result()
except Exception as e:
print(str(e))
input('\n finished……')

[pengyouak]

视频尺寸超过了GitHub的附件限制，我需要怎么发上来呢？
How can I upload a video that exceeds the attachment limit on GitHub?

[Breakthrough]

视频尺寸超过了GitHub的附件限制，我需要怎么发上来呢？ How can I upload a video that exceeds the attachment limit on GitHub?

I don't think you can, how large is the file? You may have to upload elsewhere, you may also try the Discord or upload to Google Drive.

[pengyouak]

Microsoft OneDrive 试试这个，一共74MB

[Breakthrough]

Thanks, yes this is a bug and will be fixed per #455. Note that you can use different backend like PyAV or MoviePy to avoid this issue in the meantime, thanks!

[pengyouak]

非常感谢，我试试PyAV，看起来也不错