{"ID":6620687,"CreatedAt":"2026-07-15T01:01:48.440468303Z","UpdatedAt":"2026-07-15T03:28:55.185153975Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.12786","arxiv_id":"2607.12786","title":"CoRe: A Comprehensive Framework for Cross-Image Comparative Reasoning in Vision-Language Models","abstract":"Cross-image comparative reasoning remains challenging for vision-language models (VLMs), especially when correct prediction requires fine-grained attribute grounding and globally consistent reasoning. We present CoRe, a unified framework for this problem. CoRe includes: (i) CoRe-20K, a large-scale triplet-based training set automatically constructed from structured visual metadata through a multi-expert collaborative pipeline, covering counting, depth, distance, and spatial relations; (ii) TriSR, a structured reward framework that jointly supervises attribute grounding, judgment alignment, and triplet consistency under GRPO optimization; and (iii) CoRe-Bench, the first benchmark dedicated to fine-grained cross-image comparative reasoning. Experiments show that CoRe substantially outperforms existing VLMs on CoRe-Bench while remaining competitive on standard multimodal benchmarks, achieving a 28.2-point gain in partial accuracy over the strongest baseline.","short_abstract":"Cross-image comparative reasoning remains challenging for vision-language models (VLMs), especially when correct prediction requires fine-grained attribute grounding and globally consistent reasoning. We present CoRe, a unified framework for this problem. CoRe includes: (i) CoRe-20K, a large-scale triplet-based trainin...","url_abs":"https://arxiv.org/abs/2607.12786","url_pdf":"https://arxiv.org/pdf/2607.12786v1","authors":"[\"Lin Peng\",\"Cong Wan\",\"Zeyu Guo\",\"SongLin Dong\",\"Yihong Gong\"]","published":"2026-07-14T14:01:07Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Language Model\"]","has_code":false}