Accurate aesthetic image cropping improves with continuous preference modeling
Discrete Annotation, Continuous Preference: Rethinking Supervision for Accurate and Generalizable Aesthetic Image Cropping
Computer Vision and Pattern Recognition
Summary
Choosing the best crop in a photo is tricky because people have different tastes and existing data uses limited fixed grids to label good crops. The authors propose a new way to understand preferences as a smooth and detailed map instead of isolated points. They build a model that learns from this new map, which leads to better and more adaptable cropping results. They also fix evaluation methods to fairly compare cropping tools using their new approach.
What this means in practice
- •For photo editing software developers: Integrate the proposed continuous preference model to improve automatic image cropping features, yielding more appealing results across diverse images.$Commercial implications: Enhances photo editing tools with state-of-the-art cropping accuracy attracting users through better visual results.
- •For social media platform engineers: Use the recalibrated evaluation method and model to optimize image presentation automatically on diverse user-shared photos, improving visual appeal.
Authors
Ziqing Zhang, Xiao Liu, Kai Liu, Jianze Li, Weihang Zhang, Linghe Kong, Yulun Zhang
Abstract
Aesthetic image cropping aims to identify the optimal crop of an image in terms of aesthetics and composition. While supervision based on annotated data is fundamental, the field has been hindered by a long-standing problem: existing datasets suffer from (1) human subjectivity and (2) rigid discreteness confined to fixed sampling grids. These flawed annotations not only limit the accuracy and generalization of trained models but also severely distort fair evaluation. To overcome this, we propose to model human cropping preference as a multi-peaked, continuous, and sharp field over the crop space. We introduce the Continuous Preference Field (CPF), which recovers a dense preference landscape from discrete annotations through (1) peak clustering, (2) off-lattice refinement, (3) negative shaping, and (4) field assembly. Based on this, we train CPIC, a VLM-based cropping model optimized via GRPO with the CPF reward, which overcomes template collapse, achieving state-of-the-art performance and exceptional out-of-domain generalization. Finally, to resolve the long-standing benchmark evaluation crisis, we introduce CPICD, a comprehensive recalibration of existing ground-truth boxes. By leveraging the CPF to correct grid-bound artifacts across mainstream benchmarks, CPICD establishes a rigorous and reliable foundation for future cropping research. Extensive experiments and user studies demonstrate the superiority of our CPF, CPIC, and CPICD. Code, model, and data are available at https://github.com/zzqingz/CPIC.