Instance-Completion-and-Motion-Estimation-with-Deep-Shape-Priors-for-Autonomous-Driving-Presentation

The main contribution of this project is a novel object tracking and reconstruction approach that combines learned shape priors with Gauss-Newton optimization for robust dynamic perception in autonomous driving.

Paper Style Report

Joint Optimization Notes

Tracking as a factorgraph notes

PIPEPLINE

RESULTS

Cleaner code: Coming Soon!