VER-Bench: Evaluating MLLMs on Reasoning with Fine-Grained Visual Evidence

TL;DR

VER-Bench is a new evaluation framework designed to assess multi-modal large language models' ability to identify and reason with subtle, fine-grained visual clues that occupy minimal image area, emphasizing complex reasoning over basic perception.

Contribution

The paper introduces VER-Bench, a comprehensive benchmark with structured evidence to evaluate MLLMs' skills in fine-grained visual clue extraction and reasoning, addressing limitations of existing benchmarks.

Findings

Current models struggle with extracting subtle visual clues.

Models have difficulty integrating fine-grained evidence for reasoning.

VER-Bench exposes gaps in models' visual understanding capabilities.

Abstract

With the rapid development of MLLMs, evaluating their visual capabilities has become increasingly crucial. Current benchmarks primarily fall into two main types: basic perception benchmarks, which focus on local details but lack deep reasoning (e.g., "what is in the image?"), and mainstream reasoning benchmarks, which concentrate on prominent image elements but may fail to assess subtle clues requiring intricate analysis. However, profound visual understanding and complex reasoning depend more on interpreting subtle, inconspicuous local details than on perceiving salient, macro-level objects. These details, though occupying minimal image area, often contain richer, more critical information for robust analysis. To bridge this gap, we introduce the VER-Bench, a novel framework to evaluate MLLMs' ability to: 1) identify fine-grained visual clues, often occupying on average just 0.25% of the image area; 2) integrate these clues with world knowledge for complex reasoning. Comprising 374 carefully designed questions across Geospatial, Temporal, Situational, Intent, System State, and Symbolic reasoning, each question in VER-Bench is accompanied by structured evidence: visual clues and question-related reasoning derived from them. VER-Bench reveals current models' limitations in extracting subtle visual evidence and constructing evidence-based arguments, highlighting the need to enhance models's capabilities in fine-grained visual evidence extraction, integration, and reasoning for genuine visual understanding and human-like analysis. Dataset and additional materials are available https://github.com/verbta/ACMMM-25-Materials.