PulseAugur
EN
LIVE 23:50:19

New frameworks decouple AI perception and reasoning for enhanced visual understanding · 6 sources tracked

By PulseAugur Editorial · [6 sources] ·

Researchers have introduced novel frameworks to enhance the fine-grained visual reasoning capabilities of vision-language models. Rule-VLN addresses the challenge of embodied AI agents prioritizing physical navigation over semantic rules by introducing a large-scale urban benchmark and a Semantic Navigation Rectification Module (SNRM) to instill safety awareness. Separately, Perceive-to-Reason (P2R) and PixelEyes propose methods to decouple perception from reasoning, improving performance on high-resolution images and reducing redundant trajectories in multi-turn visual reasoning tasks. AI

IMPACT These advancements in decoupling perception and reasoning could lead to more robust and compliant AI agents in real-world applications, particularly in navigation and complex visual tasks.

RANK_REASON The cluster consists of multiple research papers introducing new benchmarks, modules, and frameworks for visual reasoning in AI.

Read on Hugging Face Daily Papers →

AI-generated summary · Google Gemini · from 6 sources. How we write summaries →

New frameworks decouple AI perception and reasoning for enhanced visual understanding · 6 sources tracked

COVERAGE [6]

  1. arXiv cs.AI TIER_1 English(EN) · Jiawen Wen, Penglei Sun, Wenjie Zhang, Suixuan Qiu, Weisheng Xu, Xiaofei Yang, Xiaowen Chu ·

    Rule-VLN: Bridging Perception and Compliance via Semantic Reasoning and Geometric Rectification

    arXiv:2604.16993v2 Announce Type: replace Abstract: As embodied AI transitions to real-world deployment, the success of the Vision-and-Language Navigation (VLN) task tends to evolve from mere reachability to social compliance. However, current agents suffer from a "goal-driven tr…

  2. Hugging Face Daily Papers TIER_1 English(EN) ·

    Perceive-to-Reason: Decoupling Perception and Reasoning for Fine-Grained Visual Reasoning

    A unified framework named Perceive-to-Reason (P2R) is introduced that separates visual perception from reasoning in vision-language models through a two-stage process, improving fine-grained visual reasoning performance on high-resolution images.

  3. Hugging Face Daily Papers TIER_1 English(EN) ·

    PixelEyes: Decoupling Perception and Reasoning for Pinpoint Visual Evidence Seeking

    Multi-turn visual reasoning agents suffer from entangled reasoning and perception that cause redundant trajectories; PixelEyes addresses this by decoupling these processes through mask-guided search and semantic-region breadth-first search, demonstrated on a new benchmark with ex…

  4. arXiv cs.CV TIER_1 English(EN) · Dengxian Gong, Yuanzheng Wu, Haobo Yuan, Zhengdong Hu, Tao Zhang, Yikang Zhou, Shihao Chen, Quanzhu Niu, Kai Wang, Jason Li, Haochen Wang, Lu Qi, Shunping Ji, Ming-Hsuan Yang ·

    PixelEyes: Decoupling Perception and Reasoning for Pinpoint Visual Evidence Seeking

    arXiv:2607.00115v1 Announce Type: new Abstract: This paper explores multi-turn visual reasoning and observes that MLLMs repeatedly fail to localize the target, leading to long, redundant trajectories. We attribute this failure to the entanglement of reasoning and perception withi…

  5. arXiv cs.CV TIER_1 English(EN) · Hongxing Li, Xiufeng Huang, Dingming Li, Wenjing Jiang, Zixuan Wang, Haolei Xu, Hanrong Zhang, Haiwen Hong, Longtao Huang, Hui Xue, Weiming Lu, Jun Xiao, Yueting Zhuang, Yongliang Shen ·

    Perceive-to-Reason: Decoupling Perception and Reasoning for Fine-Grained Visual Reasoning

    arXiv:2607.01191v1 Announce Type: new Abstract: Fine-grained visual reasoning remains challenging for vision-language models, especially when small but critical visual cues are buried in high-resolution images. Existing approaches rely on repeated cropping or test-time visual sea…

  6. arXiv cs.CV TIER_1 English(EN) · Yongliang Shen ·

    Perceive-to-Reason: Decoupling Perception and Reasoning for Fine-Grained Visual Reasoning

    Fine-grained visual reasoning remains challenging for vision-language models, especially when small but critical visual cues are buried in high-resolution images. Existing approaches rely on repeated cropping or test-time visual search to introduce local evidence, but they typica…

RELATED ENTITIES

RELATED TOPICS