[2506.07218] Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward

Computer Science > Machine Learning arXiv:2506.07218 (cs) [Submitted on 8 Jun 2025 (v1), last revised 3 Mar 2026 (this version, v3)] Title:Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward Authors:Tong Xiao, Xin Xu, Zhenya Huang, Hongyu Gao, Quan Liu, Qi Liu, Enhong Chen View a PDF of the paper titled Perception-R1: Advancing Multimodal Reasoning Capabilities of MLLMs via Visual Perception Reward, by Tong Xiao and 6 other authors View PDF HTML (experimental) Abstract:Enhancing the multimodal reasoning capabilities of Multimodal Large Language Models (MLLMs) is a challenging task that has attracted increasing attention in the community. Recently, several studies have applied Reinforcement Learning with Verifiable Rewards (RLVR) to the multimodal domain in order to enhance the reasoning abilities of MLLMs. However, these works largely overlook the enhancement of multimodal perception capabilities in MLLMs, which serve as a core prerequisite and foundational component of complex multimodal reasoning. Through McNemar's test, we find that existing RLVR method fails to effectively enhance the multimodal perception capabilities of MLLMs, thereby limiting their further improvement in multimodal reasoning. To address this limitation, we propose Perception-R1, which introduces a novel visual perception reward that explicitly encourages MLLMs to perceive the visual content accurately, thereby can effectively incentivizing both their mult...