Saved in:
Bibliographic Details
Main Authors: Li, Jiahao, Lu, Yang, Zhang, Yachao, Xie, Yong, Wang, Fangyong, Xie, Yuan, Qu, Yanyun
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.16170
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912720687726592
author Li, Jiahao
Lu, Yang
Zhang, Yachao
Xie, Yong
Wang, Fangyong
Xie, Yuan
Qu, Yanyun
author_facet Li, Jiahao
Lu, Yang
Zhang, Yachao
Xie, Yong
Wang, Fangyong
Xie, Yuan
Qu, Yanyun
contents Open-vocabulary semantic segmentation (OVSS) employs pixel-level vision-language alignment to associate category-related prompts with corresponding pixels. A key challenge is enhancing the multimodal dense prediction capability, specifically this pixel-level multimodal alignment. Although existing methods achieve promising results by leveraging CLIP's vision-language alignment, they rarely investigate the performance boundaries of CLIP for dense prediction from an interpretability mechanisms perspective. In this work, we systematically investigate CLIP's internal mechanisms and identify a critical phenomenon: analogous to human distraction, CLIP diverts significant attention resources from target regions to irrelevant tokens. Our analysis reveals that these tokens arise from dimension-specific over-activation; filtering them enhances CLIP's dense prediction performance. Consequently, we propose ReFocusing CLIP (RF-CLIP), a training-free approach that emulates human distraction-refocusing behavior to redirect attention from distraction tokens back to target regions, thereby refining CLIP's multimodal alignment granularity. Our method achieves SOTA performance on eight benchmarks while maintaining high inference efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2511_16170
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Target Refocusing via Attention Redistribution for Open-Vocabulary Semantic Segmentation: An Explainability Perspective
Li, Jiahao
Lu, Yang
Zhang, Yachao
Xie, Yong
Wang, Fangyong
Xie, Yuan
Qu, Yanyun
Computer Vision and Pattern Recognition
Open-vocabulary semantic segmentation (OVSS) employs pixel-level vision-language alignment to associate category-related prompts with corresponding pixels. A key challenge is enhancing the multimodal dense prediction capability, specifically this pixel-level multimodal alignment. Although existing methods achieve promising results by leveraging CLIP's vision-language alignment, they rarely investigate the performance boundaries of CLIP for dense prediction from an interpretability mechanisms perspective. In this work, we systematically investigate CLIP's internal mechanisms and identify a critical phenomenon: analogous to human distraction, CLIP diverts significant attention resources from target regions to irrelevant tokens. Our analysis reveals that these tokens arise from dimension-specific over-activation; filtering them enhances CLIP's dense prediction performance. Consequently, we propose ReFocusing CLIP (RF-CLIP), a training-free approach that emulates human distraction-refocusing behavior to redirect attention from distraction tokens back to target regions, thereby refining CLIP's multimodal alignment granularity. Our method achieves SOTA performance on eight benchmarks while maintaining high inference efficiency.
title Target Refocusing via Attention Redistribution for Open-Vocabulary Semantic Segmentation: An Explainability Perspective
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.16170