analyze-video

Forensic video analysis automates local frame-by-frame extraction from body cam footage to flag key moments and produce chaptered timelines. Scene-aware multi-pass processing organizes frames with timestamps and metadata, generating a forensic_report.md for evidentiary review.

Yes, ffmpeg and ffprobe support scene detection in surveillance video through configurable scene-thresholds. Multi-pass extraction performs a coarse sweep followed by dense frame extraction around scene changes to identify entry and exit events while keeping all processing local.

Scene-aware frame extraction detects scene changes in evidentiary video and performs dense extraction around those transitions during multi-pass processing. It is needed when reviewing long-form surveillance or dashcam footage to locate critical events without manual scrubbing.

Forensic video analysis keeps all processing local, ensuring evidentiary video never leaves your environment. It relies on ffmpeg and ffprobe installed locally to perform frame extraction, scene detection, and timeline generation on your own hardware without uploading footage.

FFmpeg-based frame extraction depends on configurable fps, scene-threshold, and max-frames settings to balance coverage and output volume. Very high fps or low scene-thresholds on long dashcam footage can generate excessive frames and metadata, requiring careful parameter tuning to manage results.

To generate a forensic timeline, analyze surveillance video files frame-by-frame using scene-aware extraction with ffmpeg and ffprobe. The process outputs chaptered frames, metadata.json with timestamps, and a forensic_report.md containing flagged key moments for admissible litigation timelines.