Detection-Guided Keypoint Estimation for Humanoid Robots from Video Frames
Xuan Lou, Zhihuo Xu, Yuexia Wang · 2026
Read Detection-Guided Keypoint Estimation for Humanoid Robots from Video Frames by Xuan Lou, Zhihuo Xu, Yuexia Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13611796.txt.
Description
Xuan Lou, Zhihuo Xu, and Yuexia Wang built a vision system that reads the body pose of humanoid robots from plain video frames. Human pose tools transfer poorly to robots, whose proportions, stiff shells, and joints differ, and labeled robot data are scarce. Their system first finds the robot, then zooms into that region to place thirteen body points.
Heat-map outputs and smooth point decoding replaced blunt point guessing, and a skeleton check kept poses sane. Error per joint fell by nearly half against a plain baseline, and match scores beat two known pose tools tuned for the task. Human-trained tools failed badly by contrast. The authors target outside monitoring of robots at work.
Catalog record
| CREATOR | Xuan Lou, Zhihuo Xu, Yuexia Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |