Learning Color Grading, No Photo Sharing: Federated Aesthetic Preference Learning for Personalized Image Enhancement

2026-07-30

AI Summary

This work addresses personalized image enhancement by aligning with individual aesthetic preferences while preserving privacy. Traditional approaches rely on centrally collecting private photos and user ratings, posing significant privacy risks and struggling with sparse, heterogeneous feedback. To overcome these limitations, we propose FedPAIE, the first framework to integrate federated learning into this task, enabling on-device personalized color grading without uploading raw data. Our method employs a lightweight dual-cue aesthetic scorer for local calibration, which guides regularized adaptation of a CLUT-based enhancer. Notably, it operates without paired retouched data and incorporates fidelity constraints along with an excess gap penalty. Evaluated on MIT-Adobe FiveK and Flickr-AES, FedPAIE demonstrates strong open-domain personalization capability, achieving an effective balance between preference alignment and image fidelity, with only 0.293M parameters for efficient deployment.

Abstract

Personalized image enhancement should reflect individual aesthetic taste, yet learning such preferences commonly depends on private photos and ratings that are unsuitable for centralized collection. The task must infer preference from sparse, heterogeneous feedback and translate it into natural-looking color transformations on resource-constrained user devices. We introduce FedPAIE, a federated personalized aesthetic image enhancement framework for user-adaptive color grading without centralizing raw photos or ratings. FedPAIE trains a lightweight dual-cue aesthetic scorer, calibrates it into a personalized scorer on a small local support set, and freezes it to guide regularized adaptation of a lightweight CLUT enhancer from unpaired local photographs. Fidelity constraints and an excess-gap penalty regularize scorer-guided adaptation to limit proxy-score over-optimization while preserving content and natural appearance. Training remains lightweight throughout the pipeline: scorer learning updates at most 0.787M parameters, enhancer adaptation updates 0.265M, and inference retains only a 0.293M-parameter personalized enhancer. Experiments on MIT-Adobe FiveK and Flickr-AES demonstrate effective open-world personalization and a favorable balance between user preference and image fidelity. FedPAIE thus connects decentralized preference learning with efficient personalized image transformation without requiring paired user retouches.

Problem

• personalized image enhancement
• aesthetic preference learning
• federated learning
• color grading
• privacy-preserving

Innovation

• federated learning
• personalized image enhancement
• color grading
• lightweight CLUT
• aesthetic preference learning