🧤 HaWoR: World-Space Ego 3D Hands 🧤
Introducing HaWoR: a state-of-the-art approach for 3D hand motion estimation in egocentric videos, developed by Imperial College (H/T Rolandos Alexandros Potamias). Achieving new SOTA in camera pose estimation and hand motion reconstruction, with code available under Attribution-NC-ND 4.0. 💙
Key Highlights:
✅ World-space 3D hand motion estimation
✅ Novel infiller net for hand motion reconstruction
✅ Reliable even with missing frames/occlusions
✅ Robust single-shot camera trajectory estimation
📄 Paper: arxiv.org/pdf/2501.02973
🌐 Project Page: Explore here
💻 Code: Access here
👉 Stay updated on AI/ML innovations! @MachineLearning_Programming
>>Click here to continue<<