PyTorchVideo

PyTorchVideo is a deep learning library designed for video understanding tasks. It offers modular and efficient components for video classification, detection, and self-supervised learning, facilitating the development of advanced video analysis models.

Link

• Visit resource

Related resources

• spireav — graph-based Go transcoding libraryGo transcoding library to build FFmpeg filter graphs programmatically and monitor progress.
• BrightWebSocket — WebSocket client for Roku SceneGraphRFC 6455 WebSocket client library for Roku SceneGraph, running in a separate task.
• Video Seal (Meta)Meta's efficient neural video and image forensic watermarking framework with a 256-bit payload model (MIT).
• openwvOpen reimplementation of the Widevine CDM for browsers, useful for understanding and testing EME content-key handling.
• pssh-box-rs — PSSH box parser/serializer in RustRust library and CLI to parse and serialize DRM PSSH boxes (base64/hex/binary) across Widevine, PlayReady, WisePlay, Irdeto, Marlin, Nagra,…
• Using DRM with Key Rotation (Unified Streaming docs)Official technical reference on CPIX ContentKeyPeriod/ContentKeyPeriodList spec detail and VOD vs live static/dynamic CPIX origin distincti…
• Codec Wiki (x266.mov)Community-maintained wiki covering AV1/HEVC/VVC codec internals, encoder settings, and tuning guides.