Estimating a 3D Human Skeleton from a Single RGB Image by Fusing Predicted Depths from Multiple Virtual Viewpoints †
Wen-Nung Lie, Veasna Vann, Irene Amerini, Paolo Russo, Fabiana Di Ciaccio · 2024
Read Estimating a 3D Human Skeleton from a Single RGB Image by Fusing Predicted Depths from Multiple Virtual Viewpoints † by Wen-Nung Lie, Veasna Vann, Irene Amerini, Paolo Russo, Fabiana Di Ciaccio (2024) 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-11679025.txt.
Description
Wen-Nung Lie and co-authors estimate 3D human poses from a single photo in a 2024 paper. Lifting a flat image to a full skeleton is a standing vision problem, since depth is lost in the photo. Multi-view rigs solve it with many cameras, but most real settings offer one.
Their network invents extra viewpoints. One stream reads joint positions and depths from the real view, while a second guesses depths as if seen from virtual angles. A cleanup stage denoises those depths, maps them back to the original frame, and fuses all views into one skeleton. The system averaged 45.7 millimeters of joint error on standard tests. That beats prior single-view methods and rivals systems that consume whole video clips.
Catalog record
| CREATOR | Wen-Nung Lie, Veasna Vann, Irene Amerini, Paolo Russo, Fabiana Di Ciaccio |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |