Saved in:
Bibliographic Details
Main Authors: Zamini, Mohamad, Shukla, Diksha
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.24997
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908997005606912
author Zamini, Mohamad
Shukla, Diksha
author_facet Zamini, Mohamad
Shukla, Diksha
contents Open-vocabulary semantic segmentation requires assigning pixel-level semantic labels while supporting an open and unrestricted set of categories. Training-free CLIP-based approaches preserve strong zero-shot generalization but typically rely on a single inference mechanism, limiting their ability to jointly address unreliable local tokens and insufficient spatial coherence. We propose DouC, a training-free dual-branch CLIP framework that decomposes dense prediction into two complementary components. OG-CLIP improves patch-level reliability via lightweight, inference-time token gating, while FADE-CLIP injects external structural priors through proxy attention guided by frozen vision foundation models. The two branches are fused at the logit level, enabling local token reliability and structure-aware patch interactions to jointly influence final predictions, with optional instance-aware correction applied as post-processing. DouC introduces no additional learnable parameters, requires no retraining, and preserves CLIP's zero-shot generalization. Extensive experiments across eight benchmarks and multiple CLIP backbones demonstrate that DouC consistently outperforms prior training-free methods and scales favorably with model capacity.
format Preprint
id arxiv_https___arxiv_org_abs_2604_24997
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DouC: Dual-Branch CLIP for Training-Free Open-Vocabulary Segmentation
Zamini, Mohamad
Shukla, Diksha
Computer Vision and Pattern Recognition
Open-vocabulary semantic segmentation requires assigning pixel-level semantic labels while supporting an open and unrestricted set of categories. Training-free CLIP-based approaches preserve strong zero-shot generalization but typically rely on a single inference mechanism, limiting their ability to jointly address unreliable local tokens and insufficient spatial coherence. We propose DouC, a training-free dual-branch CLIP framework that decomposes dense prediction into two complementary components. OG-CLIP improves patch-level reliability via lightweight, inference-time token gating, while FADE-CLIP injects external structural priors through proxy attention guided by frozen vision foundation models. The two branches are fused at the logit level, enabling local token reliability and structure-aware patch interactions to jointly influence final predictions, with optional instance-aware correction applied as post-processing. DouC introduces no additional learnable parameters, requires no retraining, and preserves CLIP's zero-shot generalization. Extensive experiments across eight benchmarks and multiple CLIP backbones demonstrate that DouC consistently outperforms prior training-free methods and scales favorably with model capacity.
title DouC: Dual-Branch CLIP for Training-Free Open-Vocabulary Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.24997