Chu-Song Chen

A new method aims to secure semi-supervised learning against backdoor threats.

Combining SAM and MLLMs for better object localization in images.

OMTSeg advances image segmentation by combining vision and language for better object recognition.