A novel method improves the ability of Vision-Language Models to adapt to new tasks.
― 5 min read
Gangshan Wu
JointFormer enhances VOS by integrating feature extraction, matching, and memory management.
― 5 min read
SportsHHI focuses on human interactions in basketball and volleyball videos for improved analysis.
― 5 min read
A new framework enhances the adaptability of vision-language models through smart data processing.
― 6 min read
A new method improves voice separation in noisy settings with multiple speakers.
― 5 min read
Self-TPT simplifies prompt tuning for vision-language models, improving speed and efficiency.
― 7 min read