A$^2$-Edit: Precise Reference-Guided Image Editing of Arbitrary Objects and Ambiguous Masks
This work addresses the challenge of precise reference-guided image editing for arbitrary objects, which is a domain-specific problem in computer vision, and while it introduces novel techniques, it appears incremental as it builds upon existing inpainting frameworks.
The paper tackles the problem of arbitrary object image editing with coarse masks by proposing A^2-Edit, a unified inpainting framework that uses a Mixture of Transformer module and Mask Annealing Training Strategy, achieving consistent outperformance over existing approaches on benchmarks like VITON-HD and AnyInsertion.
We propose \textbf{A$^2$-Edit}, a unified inpainting framework for arbitrary object categories, which allows users to replace any target region with a reference object using only a coarse mask. To address the issues of severe homogenization and limited category coverage in existing datasets, we construct a large-scale, multi-category dataset \textbf{UniEdit-500K}, which includes 8 major categories, 209 fine-grained subcategories, and a total of 500,104 image pairs. Such rich category diversity poses new challenges for the model, requiring it to automatically learn semantic relationships and distinctions across categories. To this end, we introduce the \textbf{Mixture of Transformer} module, which performs differentiated modeling of various object categories through dynamic expert selection, and further enhances cross-category semantic transfer and generalization through collaboration among experts. In addition, we propose a \textbf{Mask Annealing Training Strategy} (MATS) that progressively relaxes mask precision during training, reducing the model's reliance on accurate masks and improving robustness across diverse editing tasks. Extensive experiments on benchmarks such as VITON-HD and AnyInsertion demonstrate that A$^2$-Edit consistently outperforms existing approaches across all metrics, providing a new and efficient solution for arbitrary object editing.