Camera-only 3D Perception

TeCo3D

A temporal representation method for camera-only 3D object detection and tracking.

Author Debao Zhang
Affiliation Beijing Institute of Technology

About

TeCo3D focuses on temporal representation learning for camera-only 3D detection and tracking in autonomous-driving scenarios, and is evaluated on the nuScenes benchmark.

Status

Manuscript in preparation. Source code is not public at this time and will be released after the paper is accepted.

View paper status →