Video Duplicate Finder is a cross-platform desktop and command-line tool that locates duplicate video and image files based on visual similarity.
The tool solves the problem of finding video duplicates that traditional file-hash approaches miss—copies with different resolutions, frame rates, watermarks, or other alterations. It works by analyzing visual content through frame comparison, with optional audio fingerprinting to detect when a shorter video is a partial clip of a longer one. An optional AI matching mode using neural image embeddings can identify heavily edited versions, crops, mirrors, zooms, and trimmed clips that pixel-based methods would overlook. All processing runs locally without external services.
The tool suits anyone managing large video libraries who needs to identify redundant files beyond exact matches. It offers multiple interfaces: a desktop GUI for Windows, Linux, and macOS; a headless CLI for scripting and automation; a web UI for remote or NAS access; and a Docker image for self-hosting. The partial clip detection feature is particularly valuable for finding scene extracts or segments ripped from longer recordings. The AI matching mode adds confidence when dealing with edited or transformed versions, though it operates as an optional second pass that only adds results rather than replacing the core visual comparison.
The project shows consistent maintenance with regular updates addressing both core functionality and user-reported issues. Development activity spans multiple areas including performance optimization, feature refinement, and cross-platform compatibility. The codebase receives attention to both the visual matching algorithms and the various interface layers, with ongoing work to improve scanning speed and detection accuracy across different video formats and transformations.
