This paper introduces a novel pipeline designed to enhance inpainting outcomes in the specific task of defurnishing, which involves the removal of furniture items from indoor panorama images. The proposed method capitalizes on Stable Diffusion, a technique that significantly improves the quality of inpainting by incorporating increased context, domain-specific model fine-tuning, and advanced image blending techniques. This approach enables the generation of high-fidelity unpaints that are not only geometrically plausible but also eliminate the necessity for room layout estimation.
The authors provide both qualitative and quantitative evidence to showcase the superiority of their method over existing furniture removal techniques. By focusing on the specific challenges associated with defurnishing tasks, such as maintaining geometric consistency and blending the unpainted areas seamlessly with the surrounding environment, this pipeline sets a new benchmark in the field. The paper is structured to ensure clarity and ease of understanding, making it accessible to both experts and novices interested in the advancements of inpainting technology and its applications in image editing.
