PosturePose: Optimized Posture Analysis for Semi-Supervised Monocular 3D Human Pose Estimation
Lawrence Amadi, Gady Agam, Irene Amerini, Paolo Russo, Fabiana Di Ciaccio · 2023
Read PosturePose: Optimized Posture Analysis for Semi-Supervised Monocular 3D Human Pose Estimation by Lawrence Amadi, Gady Agam, Irene Amerini, Paolo Russo, Fabiana Di Ciaccio (2023) 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-10747870.txt.
Description
Lawrence Amadi, Gady Agam, and co-authors present PosturePose, a method for estimating three-dimensional human poses from a single camera with little labeled data, in a 2023 paper. Labeled 3D pose sets are narrow, so the team mixes them with plentiful unlabeled video. Their key addition is a posture consistency loss that does not depend on camera angle, paired with checks on body mechanics.
Tested on two standard datasets, the approach cut pose error notably. With camera details for the unlabeled clips, it beat leading semi-supervised methods by up to fifteen percent. Without those details, gains reached seventeen percent. The authors report that their models rival other strong systems trained under the same limited-data conditions.
Catalog record
| CREATOR | Lawrence Amadi, Gady Agam, Irene Amerini, Paolo Russo, Fabiana Di Ciaccio |
|---|---|
| YEAR | 2023 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |