ViCrit: A Verifiable Reinforcement Learning Proxy Task for Visual Perception in VLMs

• URL: http://arxiv.org/abs/2506.10128v1
• Date: Wed, 11 Jun 2025 19:16:54 GMT
• Title: ViCrit: A Verifiable Reinforcement Learning Proxy Task for Visual Perception in VLMs
• Authors: Xiyao Wang, Zhengyuan Yang, Chao Feng, Yongyuan Liang, Yuhang Zhou, Xiaoyu Liu, Ziyi Zang, Ming Li, Chung-Ching Lin, Kevin Lin, Linjie Li, Furong Huang, Lijuan Wang,
• Abstract summary: We introduce ViCrit (Visual Caption Hallucination Critic), an RL proxy task that trains VLMs to localize a subtle, synthetic visual hallucination injected into paragraphs of human-written image captions.<n>Models trained with the ViCrit Task exhibit substantial gains across a variety of vision-language models benchmarks.
• Score: 98.27348724529257
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Reinforcement learning (RL) has shown great effectiveness for fine-tuning large language models (LLMs) using tasks that are challenging yet easily verifiable, such as math reasoning or code generation. However, extending this success to visual perception in vision-language models (VLMs) has been impeded by the scarcity of vision-centric tasks that are simultaneously challenging and unambiguously verifiable. To this end, we introduce ViCrit (Visual Caption Hallucination Critic), an RL proxy task that trains VLMs to localize a subtle, synthetic visual hallucination injected into paragraphs of human-written image captions. Starting from a 200-word captions, we inject a single, subtle visual description error-altering a few words on objects, attributes, counts, or spatial relations-and task the model to pinpoint the corrupted span given the image and the modified caption. This formulation preserves the full perceptual difficulty while providing a binary, exact-match reward that is easy to compute and unambiguous. Models trained with the ViCrit Task exhibit substantial gains across a variety of VL benchmarks. Crucially, the improvements transfer beyond natural-image training data to abstract image reasoning and visual math, showing promises of learning to perceive rather than barely memorizing seen objects. To facilitate evaluation, we further introduce ViCrit-Bench, a category-balanced diagnostic benchmark that systematically probes perception errors across diverse image domains and error types. Together, our results demonstrate that fine-grained hallucination criticism is an effective and generalizable objective for enhancing visual perception in VLMs.

Related papers

• CLIP-IN: Enhancing Fine-Grained Visual Understanding in CLIP via Instruction Editing Data and Long Captions [17.05291662808873]
  We present CLIP-IN, a novel framework that bolsters CLIP's fine-grained perception through two core innovations.<n> Firstly, we leverage instruction-editing datasets, originally designed for image manipulation, as a unique source of hard negative image-text pairs.<n> Secondly, CLIP-IN incorporates long captions, utilizing rotary positional encodings to capture rich semantic context often missed by standard CLIP.
  arXiv  Detail & Related papers  (2025-08-04T11:57:10Z)
• SECOND: Mitigating Perceptual Hallucination in Vision-Language Models via Selective and Contrastive Decoding [5.976839106353883]
  SECOND: Selective and Contrastive Decoding is a novel approach that enables Vision-Language Models to leverage multi-scale visual information with an object-centric manner.<n> SECOND significantly reduces perceptual hallucinations and outperforms a wide range of benchmarks.
  arXiv  Detail & Related papers  (2025-06-10T02:55:38Z)
• Towards a Systematic Evaluation of Hallucinations in Large-Vision Language Models [57.58426038241812]
  Large Vision-Language Models (LVLMs) have demonstrated remarkable performance in complex multimodal tasks.<n>These models still suffer from hallucinations when required to implicitly recognize or infer diverse visual entities from images.<n>We propose a novel visual question answering (VQA) benchmark that employs contextual reasoning prompts as hallucination attacks.
  arXiv  Detail & Related papers  (2024-12-29T23:56:01Z)
• FiVL: A Framework for Improved Vision-Language Alignment through the Lens of Training, Evaluation and Explainability [10.184567639685321]
  We introduce FiVL, a novel method for constructing datasets designed to train LVLMs for enhanced visual grounding.<n>We present benchmarks to assess the model's ability to use image as substantive evidence.<n>We identify attention heads with the strongest vision-language alignment, enabling explainability on visual-driven hallucinations.
  arXiv  Detail & Related papers  (2024-12-19T09:24:10Z)
• HELPD: Mitigating Hallucination of LVLMs by Hierarchical Feedback Learning with Vision-enhanced Penalty Decoding [36.360171373963716]
  Large Vision-Language Models (LVLMs) have shown remarkable performance on many visual-language tasks. These models still suffer from multimodal hallucination, which means the generation of objects or content that violates the images. We propose Hierarchical Feedback Learning with Vision-enhanced Penalty Decoding (HELPD) to address this issue.
  arXiv  Detail & Related papers  (2024-09-30T15:52:05Z)
• ViGoR: Improving Visual Grounding of Large Vision Language Models with Fine-Grained Reward Modeling [35.098725056881655]
  Large vision language models (LVLMs) have shown unprecedented visual reasoning capabilities. The generated text often suffers from inaccurate grounding in the visual input, resulting in errors such as hallucination of nonexistent scene elements. We introduce a novel framework, ViGoR, that utilizes fine-grained reward modeling to significantly enhance the visual grounding of LVLMs over pre-trained baselines.
  arXiv  Detail & Related papers  (2024-02-09T01:00:14Z)
• UniFine: A Unified and Fine-grained Approach for Zero-shot Vision-Language Understanding [88.24517460894634]
  We propose a unified framework to take advantage of the fine-grained information for zero-shot vision-language learning.<n>Our framework outperforms former zero-shot methods on VQA and achieves substantial improvement on SNLI-VE and VCR.
  arXiv  Detail & Related papers  (2023-07-03T09:03:12Z)
• SgVA-CLIP: Semantic-guided Visual Adapting of Vision-Language Models for Few-shot Image Classification [84.05253637260743]
  We propose a new framework, named Semantic-guided Visual Adapting (SgVA), to extend vision-language pre-trained models. SgVA produces discriminative task-specific visual features by comprehensively using a vision-specific contrastive loss, a cross-modal contrastive loss, and an implicit knowledge distillation. State-of-the-art results on 13 datasets demonstrate that the adapted visual features can well complement the cross-modal features to improve few-shot image classification.
  arXiv  Detail & Related papers  (2022-11-28T14:58:15Z)
• DiMBERT: Learning Vision-Language Grounded Representations with Disentangled Multimodal-Attention [101.99313208598569]
  Vision-and-language (V-L) tasks require the system to understand both vision content and natural language. We propose DiMBERT (short for Disentangled Multimodal-Attention BERT), which applies separated attention spaces for vision and language. We show that DiMBERT sets new state-of-the-art performance on three tasks.
  arXiv  Detail & Related papers  (2022-10-28T23:00:40Z)
• Exploring CLIP for Assessing the Look and Feel of Images [87.97623543523858]
  We introduce Contrastive Language-Image Pre-training (CLIP) models for assessing both the quality perception (look) and abstract perception (feel) of images in a zero-shot manner. Our results show that CLIP captures meaningful priors that generalize well to different perceptual assessments.
  arXiv  Detail & Related papers  (2022-07-25T17:58:16Z)

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.