Making Image Editing Easier via Adaptive Task Reformulation with Agentic Executions
Nanjing University（南京大学）；Beijing Institute of Technology（北京理工大学）；College of Artificial Intelligence, China University of Petroleum (Beijing)（中国石油大学（北京）人工智能学院）；Beijing University of Posts and Telecommunications（北京邮电大学）；Fudan University（复旦大学）
Bo Zhao, Kairui Guo, Runnan Du, Haiyang Sun, Pengshan Wang, Huan Yang, Kun Gai, Yixin Cao, Wei Ji
Instruction guided image editing has advanced substantially with recent generative models, yet it still fails to produce reliable results across many seemingly simple cases. We observe that a large portion of these failures stem not from insufficient model capacity, but from poorly formulated editing tasks, such as those involving small targets, implicit spatial relations, or under-specified instructions. In this work, we frame image editing failures as a task formulation problem and propose an adaptive task reformulation framework that improves editing performance without modifying the underlying model. Our key idea is to transform the original image-instruction pair into a sequence of operations that are dynamically determined and executed by a MLLM agent through analysis, routing, reformulation, and feedback-driven refinement. Experiments on multiple benchmarks, including ImgEdit, PICA, and RePlan, across diverse editing backbones such as Qwen Image Edit and Nano Banana, show consistent improvements, with especially large gains on challenging cases. These results suggest that task reformulation is a critical but underexplored factor, and that substantial gains can be achieved by better matching editing tasks to the effective operating regime of existing models.
Can VLMs Predict Future States? Bootstrapping World Models from Inverse Dynamics
Institute for Language, Cognition and Computation, University of Edinburgh（语言、认知与计算研究所，爱丁堡大学）；Language Technology Lab, University of Cambridge（语言技术实验室，剑桥大学）；NVIDIA（NVIDIA公司）；University of Groningen（格罗宁根大学）
