Decoupling What to Count and Where to See for Referring Expression Counting

• URL: http://arxiv.org/abs/2510.24374v1
• Date: Tue, 28 Oct 2025 12:51:53 GMT
• Title: Decoupling What to Count and Where to See for Referring Expression Counting
• Authors: Yuda Zou, Zijian Zhang, Yongchao Xu,
• Abstract summary: We propose W2-Net, a novel framework that explicitly decouples the problem into "what to count" and "where to see"<n>We introduce Subclass Separable Matching (SSM), a novel matching strategy that incorporates a repulsive force to enhance inter-subclass separability during label assignment.<n>W2-Net significantly outperforms the state-of-the-art on the REC-8K dataset, reducing counting error by 22.5% and 18.0% (test), and improving localization F1 by 7% and 8%, respectively.
• Score: 13.74402820114063
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Referring Expression Counting (REC) extends class-level object counting to the fine-grained subclass-level, aiming to enumerate objects matching a textual expression that specifies both the class and distinguishing attribute. A fundamental challenge, however, has been overlooked: annotation points are typically placed on class-representative locations (e.g., heads), forcing models to focus on class-level features while neglecting attribute information from other visual regions (e.g., legs for "walking"). To address this, we propose W2-Net, a novel framework that explicitly decouples the problem into "what to count" and "where to see" via a dual-query mechanism. Specifically, alongside the standard what-to-count (w2c) queries that localize the object, we introduce dedicated where-to-see (w2s) queries. The w2s queries are guided to seek and extract features from attribute-specific visual regions, enabling precise subclass discrimination. Furthermore, we introduce Subclass Separable Matching (SSM), a novel matching strategy that incorporates a repulsive force to enhance inter-subclass separability during label assignment. W2-Net significantly outperforms the state-of-the-art on the REC-8K dataset, reducing counting error by 22.5% (validation) and 18.0% (test), and improving localization F1 by 7% and 8%, respectively. Code will be available.

Related papers

• Dual Prompt Learning for Adapting Vision-Language Models to Downstream Image-Text Retrieval [23.472806734625774]
  We propose Dual prompt Learning with Joint Category-Attribute Reweighting (DCAR) to achieve precise image-text matching.<n>Based on the prompt paradigm, DCAR jointly optimize attribute and class features to enhance fine-grained representation learning.
  arXiv  Detail & Related papers  (2025-08-06T02:44:08Z)
• LIRA: Inferring Segmentation in Large Multi-modal Models with Local Interleaved Region Assistance [54.683384204063934]
  Large multi-modal models (LMMs) struggle with inaccurate segmentation and hallucinated comprehension.<n>We propose LIRA, a framework that capitalizes on the complementary relationship between visual comprehension and segmentation.<n>LIRA achieves state-of-the-art performance in both segmentation and comprehension tasks.
  arXiv  Detail & Related papers  (2025-07-08T07:46:26Z)
• Split Matching for Inductive Zero-shot Semantic Segmentation [56.47556212515178]
  Zero-shot Semantic (ZSS) aims to segment categories that are not annotated during training.<n>We propose Split Matching (SM), a novel assignment strategy that decouples Hungarian matching into two components.<n>SM is the first to introduce decoupled Hungarian matching under the inductive ZSS setting, and achieves state-of-the-art performance on two standard benchmarks.
  arXiv  Detail & Related papers  (2025-05-08T07:56:30Z)
• African or European Swallow? Benchmarking Large Vision-Language Models for Fine-Grained Object Classification [53.89380284760555]
  textttFOCI (textbfFine-grained textbfObject textbfClasstextbfIfication) is a difficult multiple-choice benchmark for fine-grained object classification. textttFOCIxspace complements five popular classification datasets with four domain-specific subsets from ImageNet-21k.
  arXiv  Detail & Related papers  (2024-06-20T16:59:39Z)
• DeLR: Active Learning for Detection with Decoupled Localization and Recognition Query [53.54802901197267]
  In this paper, we rethink two key components, i.e., localization and recognition, for object detection. Motivated by this, we propose an efficient query strategy, called Decoupling the localization and recognition for active query.
  arXiv  Detail & Related papers  (2023-12-28T09:58:32Z)
• Dual Feature Augmentation Network for Generalized Zero-shot Learning [14.410978100610489]
  Zero-shot learning (ZSL) aims to infer novel classes without training samples by transferring knowledge from seen classes. Existing embedding-based approaches for ZSL typically employ attention mechanisms to locate attributes on an image. We propose a novel Dual Feature Augmentation Network (DFAN), which comprises two feature augmentation modules.
  arXiv  Detail & Related papers  (2023-09-25T02:37:52Z)
• Visual Recognition by Request [111.94887516317735]
  We present a novel protocol of annotation and evaluation for visual recognition. It does not require the labeler/algorithm to annotate/recognize all targets (objects, parts, etc.) at once, but instead raises a number of recognition instructions and the algorithm recognizes targets by request. We evaluate the recognition system on two mixed-annotated datasets, CPP and ADE20K, and demonstrate its promising ability of learning from partially labeled data.
  arXiv  Detail & Related papers  (2022-07-28T16:55:11Z)
• CAD: Co-Adapting Discriminative Features for Improved Few-Shot Classification [11.894289991529496]
  Few-shot classification is a challenging problem that aims to learn a model that can adapt to unseen classes given a few labeled samples. Recent approaches pre-train a feature extractor, and then fine-tune for episodic meta-learning. We propose a strategy to cross-attend and re-weight discriminative features for few-shot classification.
  arXiv  Detail & Related papers  (2022-03-25T06:14:51Z)
• The Overlooked Classifier in Human-Object Interaction Recognition [82.20671129356037]
  We encode the semantic correlation among classes into the classification head by initializing the weights with language embeddings of HOIs. We propose a new loss named LSE-Sign to enhance multi-label learning on a long-tailed dataset. Our simple yet effective method enables detection-free HOI classification, outperforming the state-of-the-arts that require object detection and human pose by a clear margin.
  arXiv  Detail & Related papers  (2022-03-10T23:35:00Z)

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.