RefAM: Attention Magnets for Zero-Shot Referral Segmentation

• URL: http://arxiv.org/abs/2509.22650v1
• Date: Fri, 26 Sep 2025 17:59:57 GMT
• Title: RefAM: Attention Magnets for Zero-Shot Referral Segmentation
• Authors: Anna Kukleva, Enis Simsar, Alessio Tonioni, Muhammad Ferjad Naeem, Federico Tombari, Jan Eric Lenssen, Bernt Schiele,
• Abstract summary: We introduce a new method that exploits features, attention scores, from diffusion transformers for downstream tasks.<n>Key insight is that stop words act as attention magnets.<n>We propose an attention redistribution strategy, where appended stop words partition background activations into smaller clusters.
• Score: 103.98022860792504
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Most existing approaches to referring segmentation achieve strong performance only through fine-tuning or by composing multiple pre-trained models, often at the cost of additional training and architectural modifications. Meanwhile, large-scale generative diffusion models encode rich semantic information, making them attractive as general-purpose feature extractors. In this work, we introduce a new method that directly exploits features, attention scores, from diffusion transformers for downstream tasks, requiring neither architectural modifications nor additional training. To systematically evaluate these features, we extend benchmarks with vision-language grounding tasks spanning both images and videos. Our key insight is that stop words act as attention magnets: they accumulate surplus attention and can be filtered to reduce noise. Moreover, we identify global attention sinks (GAS) emerging in deeper layers and show that they can be safely suppressed or redirected onto auxiliary tokens, leading to sharper and more accurate grounding maps. We further propose an attention redistribution strategy, where appended stop words partition background activations into smaller clusters, yielding sharper and more localized heatmaps. Building on these findings, we develop RefAM, a simple training-free grounding framework that combines cross-attention maps, GAS handling, and redistribution. Across zero-shot referring image and video segmentation benchmarks, our approach consistently outperforms prior methods, establishing a new state of the art without fine-tuning or additional components.

Related papers

• Artifacts and Attention Sinks: Structured Approximations for Efficient Vision Transformers [8.486148475471271]
  Vision transformers have emerged as a powerful tool across a wide range of applications, yet their inner workings remain only partially understood.<n>We examine the phenomenon of massive tokens - tokens with exceptionally high activation norms that act as attention sinks - and artifact tokens that emerge as a byproduct during inference.<n>We introduce Fast Nystr"om Attention (FNA), a training-free method that approximates self-attention in linear time and space.
  arXiv  Detail & Related papers  (2025-07-21T19:29:03Z)
• Integrated Image Reconstruction and Target Recognition based on Deep Learning Technique [3.3410072288157155]
  We present Att-ClassiGAN, which significantly reduces the reconstruction time compared to traditional CMI approaches.<n>It delivers improved Normalized Mean Squared Error (NMSE), higher Structural Similarity Index (SSIM) and better classification outcomes for the reconstructed targets.
  arXiv  Detail & Related papers  (2025-05-07T22:34:32Z)
• Any Image Restoration via Efficient Spatial-Frequency Degradation Adaptation [158.37640586809187]
  Restoring any degraded image efficiently via just one model has become increasingly significant.<n>Our approach, termed AnyIR, takes a unified path that leverages inherent similarity across various degradations.<n>To fuse the degradation awareness and the contextualized attention, a spatial-frequency parallel fusion strategy is proposed.
  arXiv  Detail & Related papers  (2025-04-19T09:54:46Z)
• Exploring Local Memorization in Diffusion Models via Bright Ending Attention [62.979954692036685]
  "bright ending" (BE) anomaly in text-to-image diffusion models prone to memorizing training images.<n>We propose a simple yet effective method to integrate BE into existing frameworks.
  arXiv  Detail & Related papers  (2024-10-29T02:16:01Z)
• iSeg: An Iterative Refinement-based Framework for Training-free Segmentation [85.58324416386375]
  We present a deep experimental analysis on iteratively refining cross-attention map with self-attention map. We propose an effective iterative refinement framework for training-free segmentation, named iSeg. Our proposed iSeg achieves an absolute gain of 3.8% in terms of mIoU compared to the best existing training-free approach in literature.
  arXiv  Detail & Related papers  (2024-09-05T03:07:26Z)
• Spatial Structure Constraints for Weakly Supervised Semantic Segmentation [100.0316479167605]
  A class activation map (CAM) can only locate the most discriminative part of objects. We propose spatial structure constraints (SSC) for weakly supervised semantic segmentation to alleviate the unwanted object over-activation of attention expansion. Our approach achieves 72.7% and 47.0% mIoU on the PASCAL VOC 2012 and COCO datasets, respectively.
  arXiv  Detail & Related papers  (2024-01-20T05:25:25Z)
• Calibrating Undisciplined Over-Smoothing in Transformer for Weakly Supervised Semantic Segmentation [51.14107156747967]
  Weakly supervised semantic segmentation (WSSS) has attracted considerable attention because it requires fewer annotations than fully supervised approaches.<n>We propose an Adaptive Re-Activation Mechanism (AReAM) to control deep-level attention to undisciplined over-smoothing.<n>AReAM substantially improves segmentation performance compared with existing WSSS methods, reducing noise while sharpening focus on relevant semantic regions.
  arXiv  Detail & Related papers  (2023-05-04T19:11:33Z)
• Fine-Grained Attention for Weakly Supervised Object Localization [1.490944787606832]
  We propose a novel residual fine-grained attention (RFGA) module that autonomously excites the less activated regions of an object. We devise a series of mechanisms of triple-view attention representation, attention expansion, and feature calibration. We validated the superiority of our proposed RFGA module by comparing it with the recent methods in the literature over three datasets.
  arXiv  Detail & Related papers  (2021-04-11T08:14:05Z)
• ADRN: Attention-based Deep Residual Network for Hyperspectral Image Denoising [52.01041506447195]
  We propose an attention-based deep residual network to learn a mapping from noisy HSI to the clean one. Experimental results demonstrate that our proposed ADRN scheme outperforms the state-of-the-art methods both in quantitative and visual evaluations.
  arXiv  Detail & Related papers  (2020-03-04T08:36:27Z)
• Hybrid Multiple Attention Network for Semantic Segmentation in Aerial Images [24.35779077001839]
  We propose a novel attention-based framework named Hybrid Multiple Attention Network (HMANet) to adaptively capture global correlations. We introduce a simple yet effective region shuffle attention (RSA) module to reduce feature redundant and improve the efficiency of self-attention mechanism.
  arXiv  Detail & Related papers  (2020-01-09T07:47:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.