Search papers, labs, and topics across Lattice.
The paper introduces Region-to-Image Distillation, a method to improve fine-grained multimodal perception in MLLMs by distilling knowledge from zoomed-in regions to the full image during training. This approach generates high-quality VQA data from micro-cropped regions using strong teacher models and then transfers this region-grounded supervision to a student model. The resulting student model exhibits improved fine-grained perception in a single forward pass, eliminating the need for iterative zooming during inference.
Ditch the slow, iterative zooming during MLLM inference: Region-to-Image Distillation lets you bake those agentic zooming benefits directly into a single forward pass.
Multimodal Large Language Models (MLLMs) excel at broad visual understanding but still struggle with fine-grained perception, where decisive evidence is small and easily overwhelmed by global context. Recent"Thinking-with-Images"methods alleviate this by iteratively zooming in and out regions of interest during inference, but incur high latency due to repeated tool calls and visual re-encoding. To address this, we propose Region-to-Image Distillation, which transforms zooming from an inference-time tool into a training-time primitive, thereby internalizing the benefits of agentic zooming into a single forward pass of an MLLM. In particular, we first zoom in to micro-cropped regions to let strong teacher models generate high-quality VQA data, and then distill this region-grounded supervision back to the full image. After training on such data, the smaller student model improves"single-glance"fine-grained perception without tool use. To rigorously evaluate this capability, we further present ZoomBench, a hybrid-annotated benchmark of 845 VQA data spanning six fine-grained perceptual dimensions, together with a dual-view protocol that quantifies the global--regional"zooming gap". Experiments show that our models achieve leading performance across multiple fine-grained perception benchmarks, and also improve general multimodal cognition on benchmarks such as visual reasoning and GUI agents. We further discuss when"Thinking-with-Images"is necessary versus when its gains can be distilled into a single forward pass. Our code is available at https://github.com/inclusionAI/Zooming-without-Zooming.
