[2512.03992] Value-Guided Iterative Refinement and the DIQ-H Benchmark for Evaluating VLM Robustness

Computer Science > Computer Vision and Pattern Recognition arXiv:2512.03992 (cs) [Submitted on 3 Dec 2025 (v1), last revised 29 Apr 2026 (this version, v2)] Title:Value-Guided Iterative Refinement and the DIQ-H Benchmark for Evaluating VLM Robustness Authors:Hanwen Wan, Zexin Lin, Yixuan Deng, Xiaoqiang Ji View a PDF of the paper titled Value-Guided Iterative Refinement and the DIQ-H Benchmark for Evaluating VLM Robustness, by Hanwen Wan and 3 other authors View PDF HTML (experimental) Abstract:Vision-Language Models (VLMs) are essential for embodied AI and safety-critical applications, such as robotics and autonomous systems. However, existing benchmarks primarily focus on static or curated visual inputs, neglecting the challenges posed by adversarial conditions, value misalignment, and error propagation in continuous deployment. Current benchmarks either overlook the impact of real-world perturbations, or fail to account for the cumulative effect of inconsistent reasoning over time. To address these gaps, we introduce the Degraded Image Quality Leading to Hallucinations (DIQ-H) benchmark, the first to evaluate VLMs under adversarial visual conditions in continuous sequences. DIQ-H simulates real-world stressors including motion blur, sensor noise, and compression artifacts, and measures how these corruptions lead to persistent errors and misaligned outputs across time. The benchmark explicitly models error propagation and its long-term value consistency. To enhance scala...