From Pixels to Concepts: Do Segmentation Models Understand What They Segment?

• URL: http://arxiv.org/abs/2605.09591v1
• Date: Sun, 10 May 2026 15:07:17 GMT
• Title: From Pixels to Concepts: Do Segmentation Models Understand What They Segment?
• Abstract summary: We introduce CAFE: textbfCounterfactual textbfAttribute textbfFactuality textbfEvaluation, a novel benchmark for evaluating concept-faithful segmentation in promptable segmentation models.<n>Our benchmark contains 2,146 paired test samples, each consisting of a target image, a ground-truth mask, a positive prompt, and a misleading negative prompt.
• Score: 26.89265297426196
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Segmentation is a fundamental vision task underlying numerous downstream applications. Recent promptable segmentation models, such as Segment Anything Model 3 (SAM3), extend segmentation from category-agnostic mask prediction to concept-guided localization conditioned on high-level textual prompts. However, existing benchmarks primarily evaluate mask accuracy or object presence, leaving unclear whether these models faithfully ground the queried concept or instead rely on visually salient but semantically misleading cues. We introduce CAFE: \textbf{C}ounterfactual \textbf{A}ttribute \textbf{F}actuality \textbf{E}valuation, a novel benchmark for evaluating concept-faithful segmentation in promptable segmentation models. Our \textbf{CAFE} is built on attribute-level counterfactual manipulation: the target region and ground-truth mask are preserved, while attributes such as surface appearance, context, or material composition are modified to introduce misleading semantic cues. The benchmark contains 2,146 paired test samples, each consisting of a target image, a ground-truth mask, a positive prompt, and a misleading negative prompt. These samples cover three counterfactual categories: Superficial Mimicry (\textbf{SM}), Context Conflict (\textbf{CC}), and Ontological Conflict (\textbf{OC}). We evaluate various model types and sizes on our CAFE. Experiments reveal a systematic gap between localization quality and concept discrimination: models often generate accurate masks even for misleading prompts, suggesting that strong mask prediction does not necessarily imply faithful semantic grounding. Our CAFE provides a controlled benchmark for diagnosing whether promptable segmentation models perform concept-faithful grounding rather than shortcut-driven mask retrieval.

Related papers

• Prompting Diffusion Models for Zero-Shot Instance Segmentation [74.37083707535778]
  We present Prompt2Seg, a spatial conditioning framework for diffusion-based segmentation.<n>Our approach takes spatial prompts, represented as 2D Gaussians or confidence maps, as explicit input signals, training the model to respond directly to user intent.<n>We evaluate on seven datasets ranging from standard benchmarks to more challenging domains, including paintings, egocentric views, and X-ray data.
  arXiv  Detail & Related papers  (2026-06-21T20:30:41Z)
• Boundary-by-Mask: Few-Shot Instance Segmentation with Mask-Conditioned Boundary Learning for Texture-Poor Industrial Parts [1.3873323883842137]
  We propose Boundary-by-Mask, a few-shot instance segmentation framework that supervises boundaries instead of interior appearance.<n>The framework achieves reliable instance separation even on low-uniform and color-uniform surfaces.<n>Experiments on industrial parts and food items with ambiguous boundaries show strong few-shot generalization, robustness in feature-poor conditions, and precise control over mask-level targets.
  arXiv  Detail & Related papers  (2026-06-19T16:50:28Z)
• S$^3$POT: Contrast-Driven Face Occlusion Segmentation via Self-Supervised Prompt Learning [46.05577414378133]
  We present S$3$POT, a contrast-driven framework synergizing face generation with self-supervised spatial prompting.<n>In particular, S$3$POT consists of three modules: Reference Generation, Feature enhancement, and Prompt Selection.<n>Experiments on a dedicatedly collected dataset demonstrate S$3$POT's superior performance and the effectiveness of each module.
  arXiv  Detail & Related papers  (2026-01-31T10:05:13Z)
• ResAgent: Entropy-based Prior Point Discovery and Visual Reasoning for Referring Expression Segmentation [21.87321809019825]
  Referring Expression (RES) is a core vision-language segmentation task that enables pixel-level understanding of targets via free-form linguistic expressions.<n>textbfmodel is a novel RES framework integrating textbfEntropy-textbfBased Point textbfDiscovery (textbfEBD) and textbfVision-textbfBased textbfReasoning (textbfVBR)<n>model implements a coarse-to
  arXiv  Detail & Related papers  (2026-01-23T01:56:04Z)
• SPARTA: Evaluating Reasoning Segmentation Robustness through Black-Box Adversarial Paraphrasing in Text Autoencoder Latent Space [11.534994345027362]
  Multimodal large language models (MLLMs) have shown impressive capabilities in vision-language tasks such as reasoning segmentation.<n>We introduce a novel adversarial paraphrasing task: generating grammatically correct paraphrases that preserve the original query meaning while degrading segmentation performance.<n>We introduce SPARTA-a black-box, sentence-level optimization method that operates in the low-dimensional semantic latent space of a text autoencoder.
  arXiv  Detail & Related papers  (2025-10-28T14:09:05Z)
• LESS: Label-Efficient and Single-Stage Referring 3D Segmentation [55.06002976797879]
  Referring 3D is a visual-language task that segments all points of the specified object from a 3D point cloud described by a sentence of query. We propose a novel Referring 3D pipeline, Label-Efficient and Single-Stage, dubbed LESS, which is only under the supervision of efficient binary mask. We achieve state-of-the-art performance on ScanRefer dataset by surpassing the previous methods about 3.7% mIoU using only binary labels.
  arXiv  Detail & Related papers  (2024-10-17T07:47:41Z)
• Bridge the Points: Graph-based Few-shot Segment Anything Semantically [79.1519244940518]
  Recent advancements in pre-training techniques have enhanced the capabilities of vision foundation models. Recent studies extend the SAM to Few-shot Semantic segmentation (FSS) We propose a simple yet effective approach based on graph analysis.
  arXiv  Detail & Related papers  (2024-10-09T15:02:28Z)
• Boosting Unsupervised Semantic Segmentation with Principal Mask Proposals [15.258631373740686]
  Unsupervised semantic segmentation aims to automatically partition images into semantically meaningful regions by identifying global semantic categories within an image corpus without any form of annotation. We present PriMaPs - Principal Mask Proposals - decomposing images into semantically meaningful masks based on their feature representation. This allows us to realize unsupervised semantic segmentation by fitting class prototypes to PriMaPs with a expectation-maximization algorithm, PriMaPs-EM.
  arXiv  Detail & Related papers  (2024-04-25T17:58:09Z)
• CPCM: Contextual Point Cloud Modeling for Weakly-supervised Point Cloud Semantic Segmentation [60.0893353960514]
  We study the task of weakly-supervised point cloud semantic segmentation with sparse annotations. We propose a Contextual Point Cloud Modeling ( CPCM) method that consists of two parts: a region-wise masking (RegionMask) strategy and a contextual masked training (CMT) method.
  arXiv  Detail & Related papers  (2023-07-19T04:41:18Z)
• Reflection Invariance Learning for Few-shot Semantic Segmentation [53.20466630330429]
  Few-shot semantic segmentation (FSS) aims to segment objects of unseen classes in query images with only a few annotated support images. This paper proposes a fresh few-shot segmentation framework to mine the reflection invariance in a multi-view matching manner. Experiments on both PASCAL-$5textiti$ and COCO-$20textiti$ datasets demonstrate the effectiveness of our approach.
  arXiv  Detail & Related papers  (2023-06-01T15:14:58Z)
• Location-Aware Self-Supervised Transformers [74.76585889813207]
  We propose to pretrain networks for semantic segmentation by predicting the relative location of image parts. We control the difficulty of the task by masking a subset of the reference patch features visible to those of the query. Our experiments show that this location-aware pretraining leads to representations that transfer competitively to several challenging semantic segmentation benchmarks.
  arXiv  Detail & Related papers  (2022-12-05T16:24:29Z)
• Discovering Object Masks with Transformers for Unsupervised Semantic Segmentation [75.00151934315967]
  MaskDistill is a novel framework for unsupervised semantic segmentation. Our framework does not latch onto low-level image cues and is not limited to object-centric datasets.
  arXiv  Detail & Related papers  (2022-06-13T17:59:43Z)
• Per-Pixel Classification is Not All You Need for Semantic Segmentation [184.2905747595058]
  Mask classification is sufficiently general to solve both semantic- and instance-level segmentation tasks. We propose MaskFormer, a simple mask classification model which predicts a set of binary masks. Our method outperforms both current state-of-the-art semantic (55.6 mIoU on ADE20K) and panoptic segmentation (52.7 PQ on COCO) models.
  arXiv  Detail & Related papers  (2021-07-13T17:59:50Z)

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.