AI RESEARCH
Self-Calibrated CLIP for Training-Free Open-Vocabulary Segmentation
arXiv CS.CV
•
ArXi:2411.15869v3 Announce Type: replace Recent advancements in pre-trained vision-language models like CLIP have enabled the task of open-vocabulary segmentation. CLIP nstrates impressive zero-shot capabilities in various downstream tasks that require holistic image understanding. However, due to the image-level contrastive learning and fully global feature interaction, ViT-based CLIP struggles to capture local details, resulting in poor performance in segmentation tasks.