# Discrete Annotation, Continuous Preference: Rethinking Supervision for Accurate and Generalizable Aesthetic Image Cropping

arXiv: https://arxiv.org/abs/2610.00582 · HTML: https://mlsanity.com/paper/2610.00582/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 24 of its 37 references that are in the map.
- First posted on arXiv 30 Sep 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 24 of 24)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [LGM: Large Multi-View Gaussian Model for High-Resolution 3D Content Creation](https://mlsanity.com/paper/2402.05054/) (2024)
- [Deep Image Blending](https://mlsanity.com/paper/1910.11495/) (2019)
- [Reliable and Efficient Image Cropping: A Grid Anchor based Approach](https://mlsanity.com/paper/1904.04441/) (2019)
- [Quantitative Analysis of Automatic Image Cropping Algorithms: A Dataset and Comparative Study](https://mlsanity.com/paper/1701.01480/) (2017)
- [Grid Anchor based Image Cropping: A New Benchmark and An Efficient Model](https://mlsanity.com/paper/1909.08989/) (2019)
- [Learning to Compose with Professional Photographs on the Web](https://mlsanity.com/paper/1702.00503/) (2017)
- [Image Composition Assessment with Saliency-augmented Multi-pattern Pooling](https://mlsanity.com/paper/2104.03133/) (2021)
- [Learning Subject-Aware Cropping by Outpainting Professional Photos](https://mlsanity.com/paper/2312.12080/) (2023)
- [Human-centric Image Cropping with Partition-aware and Content-preserving Features](https://mlsanity.com/paper/2207.10269/) (2022)
- [Cropper: Vision-Language Model for Image Cropping through In-Context Learning](https://mlsanity.com/paper/2408.07790/) (2024)
- [Beyond Image Borders: Learning Feature Extrapolation for Unbounded Image Composition](https://mlsanity.com/paper/2309.12042/) (2023)
- [The Fourth Challenge on Image Super-Resolution (×4) at NTIRE 2026: Benchmark Results and Method Overview](https://mlsanity.com/paper/2604.14558/) (2026)
- [Dog-IQA: Standard-guided Zero-shot MLLM for Mix-grained Image Quality Assessment](https://mlsanity.com/paper/2410.02505/) (2024)
- [ProCrop: Learning Aesthetic Image Cropping from Professional Compositions](https://mlsanity.com/paper/2505.22490/) (2025)
- [The First Challenge on Remote Sensing Infrared Image Super-Resolution at NTIRE 2026: Benchmark Results and Method Overview](https://mlsanity.com/paper/2604.21312/) (2026)
- [Venus: Benchmarking and Empowering Multimodal Large Language Models for Aesthetic Guidance and Cropping](https://mlsanity.com/paper/2602.23980/) (2026)
- [Accelerating Rectified Flow Models via Trajectory-Aware Caching](https://mlsanity.com/paper/2605.16789/) (2026)
- [DRScaffold: Boosting Dense-Scene Reasoning in Lightweight Vision Language Models](https://mlsanity.com/paper/2605.26038/) (2026)
- [COMEX: A Composition-Grounded Benchmark and Learning Framework for Explainable Aesthetic Image Cropping](https://mlsanity.com/paper/2608.07570/) (2026)

## Abstract (arXiv)

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.