This is an AV (Audio-Video) processing pipeline, used to generate QC Reports for the AV files. Designed to support Zoom recordings, the pipeline can be extended to support other AV files as well.
- Supports decrytion of encrypted AV files
- Uses
cryptease
to decrypt the encrypted AV files
- Uses
- Extracts Metadata from the AV files using
ffprobe
- Detects if videos have black-bars using
opencv
- Split Interviews into Left and Right Streams using
ffmpeg
- Assumes that the video has a 2 Column layout
- Each column is a different speaker (Interviewer and Subject)
- Extracts facial features using
OpenFace
- Generates a QC Report for the AV files
- The QC Report includes the following:
- Metadata
- Facial Features
- Openface Face in Frame
- Openface Confidence
- The QC Report includes the following:
- Python
- OpenCV
- FFmpeg
- OpenFace
- Postgres (Database)