arXiv cs.CL· Haoxiang Cao, Yuxuan Zhang, Penghui Du, Bo Li, Jiajiong Cao, Yingying Fan, Zimeng Wu, Yifu Luo, Mingzhe Zheng, Junwen Miao, Huaisong Zhang, Yang Yao, Yu Wang, Dongfu Jiang, Ping Nie, Wenhu Chen, Chaoqun Wang, Changqian Yu, Kelsey R. Allen·· 11 小时前AI 评分38
LongBanana：面向长上下文多参考图像合成的专家验证基准
LongBanana: An Expert-Verified Benchmark for Long-Context Multi-Reference Image Synthesis
AI 导读
研究者提出 LongBanana，一个面向长上下文多参考图像合成的专家设计与验证基准，含 90 个任务、每任务约 32 张参考图与 20 条约束，共 4,670 个原子评估项。
正文 · 原文
Authors:Haoxiang Cao, Yuxuan Zhang, Penghui Du, Bo Li, Jiajiong Cao, Yingying Fan, Zimeng Wu, Yifu Luo, Mingzhe Zheng, Junwen Miao, Huaisong Zhang, Yang Yao, Yu Wang, Dongfu Jiang, Ping Nie, Wenhu Chen, Chaoqun Wang, Changqian Yu, Kelsey R. Allen
Abstract:We introduce LongBanana, to our knowledge the first expert-designed and expert-verified benchmark for long-context multi-reference image synthesis in the age of agent-driven image generation. The task requires retrieving dispersed visual evidence, rejecting distractors, binding entities and relations, and composing one coherent final image from multiple references. LongBanana contains 90 carefully designed tasks organized by a 3-by-3 taxonomy; each task includes about 32 reference images, 20 constraints, and 50 evaluation items, yielding 4,670 expert-specified atomic evaluation items in total. Its graph-grounded evaluation design instantiates each task with reference images and an entity-relation graph, then converts node, edge, and whole-graph obligations into checklist items for Content Fidelity, Relational Binding, and Global Constraint Satisfaction. Reviewed expert labels include a 15% double-labeled subset with 89.4% exact agreement and 0.83 quadratic weighted kappa. We further introduce LongBanana-Gen, an agent-harness framework for long-horizon image generation; experiments show that minimal agent harnesses improve over direct generation, while a ReAct-Plan-Exec-Reflect (RPR) workflow yields harness-dependent gains by strengthening evidence grounding and repair. Together, LongBanana provides an auditable testbed for measuring dense visual evidence use and a shared target for developing stronger agent-driven image-generation systems. Code and data are available at the project page: this https URL.
| Comments: | NeurIPS 2026 Evaluations and Datasets Track. Project page: this https URL |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.06856 [cs.CV] |
| (or arXiv:2611.06856v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.06856 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: HaoXiang Cao [view email]
[v1]
Wed, 4 Nov 2026 18:11:56 UTC (11,055 KB)
来源：arXiv cs.CL · arxiv.org