GMMT

Generative-based Fusion Mechanism for Multi-modal Object Tracking

This work explores the potential of fusing the multi-modal information via generative models. Conditional GAN and diffusion model are involved. The results show that the GMs can produce better performance by generating features with less noise.

🫵Find our survey work at another repo

Pipeline

Resuts on GTOT

Resuts on RGBT234 and LasHeR

Resuts on RGBD1K