Visual Evidence Under Cross-Examination: Evaluating and Controlling Decision-Level Evidence Use in Vision-Language Models
cs.CV
Submitted: 2026-10-07
Updated: 2026-10-07
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence
- Qwen3-VL Technical Report
- Qwen2.5-VL Technical Report
- CogFlow: Bridging Perception and Reasoning through Knowledge Internalization for Visual Mathematical Problem Solving
- VLM-R$^3$: Region Recognition, Reasoning, and Refinement for Enhanced Multimodal Chain-of-Thought
- Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs
- Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoning
- RegionReasoner: Region-Grounded Multi-Round Visual Reasoning
- Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality
- Traceable Evidence Enhanced Visual Grounded Reasoning: Evaluation and Methodology
- Pixel Reasoner: Incentivizing Pixel-Space Reasoning with Curiosity-Driven Reinforcement Learning
- InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency
- Compose and Fuse: Revisiting the Foundational Bottlenecks in Multimodal Reasoning
- V*: Guided Visual Search as a Core Mechanism in Multimodal LLMs
- DeFacto: Counterfactual Thinking with Images for Enforcing Evidence-Grounded and Faithful Reasoning
- Set-of-Mark Prompting Unleashes Extraordinary Visual Grounding in GPT-4V
- DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models