---
title: "Beyond Pipeline Mimicry: Expert-level Aesthetic ISP via Reward-Guided Flow"
authors:
  - "Tong Qiao"
  - "Kepeng Xu"
  - "Gang He"
  - "Zhenyang Liu"
  - "Wenxin Yu"
doi: "10.24963/ijcai.2026/174"
url: "https://ijcai.org/proceedings/2026/174"
pdf: "https://ijcai.org/proceedings/2026/0174.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "1559-1567"
track: "Main Track"
keywords:
  - "Computer Vision: Image and video retrieval"
  - "Computer Vision: Applications and Systems"
  - "Knowledge Representation and Reasoning: Applications"
---

# Beyond Pipeline Mimicry: Expert-level Aesthetic ISP via Reward-Guided Flow

Tong Qiao, Kepeng Xu, Gang He, Zhenyang Liu, Wenxin Yu

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 1559-1567. DOI: [10.24963/ijcai.2026/174](https://doi.org/10.24963/ijcai.2026/174)

## Abstract

Recent deep learning-based ISP methods are primarily constrained by mimicking fixed camera pipelines, consequently struggling to achieve expert-level aesthetic quality. To this end, we propose AesISP, the first expert-level aesthetic ISP framework formulated as a reward flow model. Addressing the ill-posed nature of ISP, AesISP establishes a Privileged Prior Distillation paradigm. By leveraging expert-image-guided latent proxies, we decompose the intractable task into a tractable, proxy-driven learning process. Subsequently, we propose MeanFlow++, which reformulates the flow matching objective via target-prediction parameterization and a Progressive Temporal Curriculum. By evolving from learning instantaneous to long-range average velocities, this mechanism rectifies transport trajectories and uses deterministic mapping to impose explicit constraints, anchoring the generation trajectory to the expert manifold. Finally, we introduce a multi-dimensional reward-driven reinforcement learning approach.Leveraging Group Relative Policy Optimization (GRPO) to balance trade-offs across dimensions such as color and lighting, it steers the model to converge precisely on expert-level aesthetic standards. Experiments demonstrate that AesISP outperforms state-of-the-art methods in both quantitative metrics and aesthetic quality.