PerceptionBench: Evaluating Atomic Visual Perception in Multimodal Large Language Models

Hugging Face Daily Papers Papers
multimodal-llm visual-perception benchmark evaluation perception-bench

Summary

PerceptionBench is a benchmark designed to evaluate atomic visual perception capabilities of Multimodal Large Language Models (MLLMs), using a bottom-up taxonomy of ten atomic perceptual capabilities. Results across 16 frontier MLLMs show no model reaches 60% accuracy, indicating visual perception remains largely unsolved.

We introduce PerceptionBench, a benchmark specifically designed to evaluate the atomic visual perception capabilities of Multimodal Large Language Models (MLLMs). Existing benchmarks often fail to isolate perception: holistic evaluations conflate perceptual errors with failures in reasoning or domain knowledge, while application-driven benchmarks only cover narrow, fragmented domains shaped by heuristic designs. To address these limitations, PerceptionBench adopts a bottom-up approach: by diagnosing the earliest failure points in the responses of frontier MLLMs across 42 existing benchmarks, we construct an error taxonomy whose perception branch defines ten atomic perceptual capabilities. Guided by this taxonomy, we construct 3,000 verified questions with short, unambiguous answers, each isolating a single capability, with difficulty stemming from perception rather than reasoning or knowledge. Benchmark results across sixteen frontier MLLMs reveal that atomic perception remains largely unsolved---no model reaches 60\% accuracy, perception-related hallucination is the weakest capability on average, and similar overall scores conceal sharply divergent capability profiles. PerceptionBench thus provides a capability-level standard for measuring and diagnosing the visual perception boundaries of MLLMs.
Original Article
View Cached Full Text

Cached at: 07/29/26, 03:50 AM

Paper page - PerceptionBench: Evaluating Atomic Visual Perception in Multimodal Large Language Models

Source: https://huggingface.co/papers/2607.24957 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

WeintroducePerceptionBench,abenchmarkspecificallydesignedtoevaluatetheatomicvisualperceptioncapabilitiesofMultimodalLargeLanguageModels(MLLMs).Existingbenchmarksoftenfailtoisolateperception:holisticevaluationsconflateperceptualerrorswithfailuresinreasoningordomainknowledge,whileapplication-drivenbenchmarksonlycovernarrow,fragmenteddomainsshapedbyheuristicdesigns.Toaddresstheselimitations,PerceptionBenchadoptsabottom-upapproach:bydiagnosingtheearliestfailurepointsintheresponsesoffrontierMLLMsacross42existingbenchmarks,weconstructanerrortaxonomywhoseperceptionbranchdefinestenatomicperceptualcapabilities.Guidedbythistaxonomy,weconstruct3,000verifiedquestionswithshort,unambiguousanswers,eachisolatingasinglecapability,withdifficultystemmingfromperceptionratherthanreasoningorknowledge.BenchmarkresultsacrosssixteenfrontierMLLMsrevealthatatomicperceptionremainslargelyunsolved---nomodelreaches60\%accuracy,perception-relatedhallucinationistheweakestcapabilityonaverage,andsimilaroverallscoresconcealsharplydivergentcapabilityprofiles.PerceptionBenchthusprovidesacapability-levelstandardformeasuringanddiagnosingthevisualperceptionboundariesofMLLMs.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2607\.24957

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.24957 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.24957 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.24957 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Can Multimodal Large Language Models Understand OCT?

Hugging Face Daily Papers

This paper introduces OCT-Bench, a comprehensive benchmark for evaluating multimodal large language models on OCT image understanding, comprising 10,076 questions across 20 fine-grained tasks in perception, cognition, and reasoning. Evaluation of 20 models shows current MLLMs fall short of reliable OCT understanding.