Shanghai AI Lab's InfiniHand Tracks 3D Hands in World Space at 11 FPS from Egocentric Video
Shanghai-AI-Laboratory · hf · 2026-09-29
Shanghai AI Lab presents InfiniHand, an end-to-end streaming feed-forward framework that jointly estimates MANO hand parameters, camera trajectories, and world-space hand locations from uncalibrated egocentric video, replacing cascaded pose-estimator+SLAM pipelines. Trained on 5,000 hours of egocentric data in two progressive stages, it cuts ARCTIC PA-p by 21.4% vs ViDiHand, mitigates world-space drift, and runs at 11.19 FPS — over 2x HaWoR's throughput.
More from Embodied
- David Patterson: humanoid robots will tunnel cities' streets underground into parks — davidpattersonx · 2026-09-29
- Reality Check launches first public robot manipulation benchmark with 14,400 real-world rollouts — DJiafei · 2026-09-29
- Inside Dexory's factory: how warehouse-scanning robots go from R&D to assembly in three stages — lukas_m_ziegler · 2026-09-29
- Deep Robotics files for STAR Market IPO to raise $350M, sold just 1 humanoid robot in 2025 — 创业邦 · 2026-09-29
- EgoDemo, an egocentric human demonstration dataset for embodied AI, trends on HF — LightwheelAI · 2026-09-29
- DroneWAM: JEPA world-action model boosts drone navigation with adaptive rollout depth — Liang Yao · 2026-09-29