MoSoLoco: Monocular 3D Human Localization through Visual Latent Features and Social Context
MoSoLoco: Monocular 3D Human Localization through Visual Latent Features and Social Context
This paper introduces MoSoLoco, a per-frame monocular 3D human localizer that combines a pose detector's internal visual features with detected 2D skeletons to estimate metric 3D positions without additional RGB or depth encoders. The method incorporates sparse same-frame multi-person context to refine individual position estimates. Experiments on JRDB and NuScenes show that reusing pose-detector representations reduces the best JRDB error by approximately 23 percent relative to the strongest evaluated keypoint-based baseline. A sparse social-interaction module incorporates nearby detections from the same image for further refinement. Controlled trajectory-forecasting experiments show downstream benefits, while the complete pipeline remains suitable for efficient Jetson AGX Xavier inference.
The authors report a method that reuses internal visual features from a pose detector to estimate metric 3D human positions from monocular images, achieving approximately 23% lower error on the JRDB benchmark compared to a keypoint-based baseline while remaining suitable for Jetson AGX Xavier inference.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
alization, reducing the best JRDB error by approximately 23 percent relative to the strongest evaluated keypoint-based baseline. A sparse social-interaction module incorporates nearby detections from the same image for further refinement. Controlled trajectory-forecasting experiments show downstream benefits, while the complete pipeline remains suitable for efficient Jetson AGX Xavier inference.
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org