Image recognition

Stress-test your off-road segmentation model: 165 labelled frames under dust, night, fog, rain and mud, plus the scoring script.

Standalone fork of FrameEye Bridge with Windows packaging and gaze/eyelid smoothing for VRCFaceTracking and direct VRChat OSC.

AI Search for Every Photo and Every Frame of Video on macOS

Kinect Depth AutoLearn is a Processing + ONNX/PyTorch pipeline for building a depth-only structured human-pose model from synchronized RGB-D Kinect data.

Real-ESRGAN aims at developing Practical Algorithms for General Image/Video Restoration.

A Three.js (TSL / WebGPU) port of OpenDLSS-NR, the open-source reimplementation of NVIDIA's DLSS 5 neural rendering network

Camera-based gesture scrolling for macOS, with local hand tracking, repeated fingertip strokes, and a browser testing lab.

Run Cloudflare's clef-flash decision model locally on your webcam

Fan-made sci-fi HUD for Rokid Glasses: face tracking, reticle, local gesture detection and fictional dossiers.

Real-time, open-source running gait and form analysis engine built with Rust, WebAssembly, and computer vision pose estimation.

Look at a screen, start typing. macOS menu bar app that moves focus and cursor to the display you're facing, using head-pose detection from your Mac's camera. 100% on-device, nothing recorded or sent.

OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction

AV1 Structure from Motion, built for fun and to test Claude's abilities

Open Med Tray: procedural medication scenes for computer-vision research