The project addresses these constraints by combining:
| Problem | Quick Fix | |---------|-----------| | | • On Linux, run ls /dev/video* . If none appear, enable the camera in BIOS or load the uvcvideo kernel module ( sudo modprobe uvcvideo ). • On Windows, check Device Manager → Imaging devices. | | ffmpeg not found / “codec not supported” | Ensure ffmpeg is on your PATH ( ffmpeg -version ). The pip wheel bundles a copy, but if you use the source build you must have the external binary. | | Video is choppy / low FPS | • Lower the resolution (e.g., 640x480 ). • Verify that the USB port is not shared with high‑bandwidth devices. • Use a dedicated USB 3.0 port for the webcam. | | Audio is missing | FaceMP4 only records audio if the device index also provides a microphone. Specify --audio-device 0 (or the appropriate index) if you have a separate mic. | | Face detection is slow | Install the optional torch‑based detector: pip install facemp4[torch] . It uses a tiny MobileNet‑SSD model that runs at >30 fps on a modern laptop GPU. | | The saved file won’t play in VLC | Re‑encode with ffmpeg: ffmpeg -i shweta_cam.mp4 -c:v libx264 -crf 23 -c:a aac -b:a 128k fixed.mp4 | kaamuk shweta cam show wid facemp4 install