AlbumFill: Album-Guided Reasoning and Retrieval for Personalized Image Completion
2026-05-04 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionInformation Retrieval
AI summaryⓘ
The authors created AlbumFill, a tool that fills in missing parts of personal photos while keeping the person's identity consistent. Unlike earlier methods that need manually chosen similar photos or struggle with identity, AlbumFill automatically finds matching images from a person's photo collection. It uses a vision-language model to understand what is missing and then searches for suitable references to guide the completion. They also made a large dataset to help study this task and showed that finding the right references is very important for good results.
personalized image completionimage inpaintingidentity consistencyvision-language modelreference-based image retrievalphoto albumsemantic cuesdataset creationhuman-centric images
Authors
Yu-Ju Tsai, Brian Price, Qing Liu, Luis Figueroa, Daniil Pakhomov, Zhihong Ding, Scott Cohen, Ming-Hsuan Yang
Abstract
Personalized image completion aims to restore occluded regions in personal photos while preserving identity and appearance. Existing methods either rely on generic inpainting models that often fail to maintain identity consistency, or assume that suitable reference images are explicitly provided. In practice, suitable references are often not explicitly provided, requiring the system to search for identity-consistent images within personal photo collections. We present AlbumFill, a training-free framework that retrieves identity-consistent references from personal albums for personalized completion. Given an occluded image and a personal album, a vision-language model infers missing semantic cues to guide composed image retrieval, and the retrieved references are used by reference-based completion models. To facilitate this task, we introduce a dataset containing 54K human-centric samples with associated album images. Experiments across multiple baselines demonstrate the difficulty of personalized completion and highlight the importance of identity-consistent reference retrieval. Project Page: https://liagm.github.io/AlbumFill/