Paper Detail
Jiyun Won, Heemin Yang, Woohyeok Kim, Jungseul Ok, Sunghyun Cho
Recent work has explored optimizing image signal processing (ISP) pipelines for various tasks by composing predefined modules and adapting them to task-specific objectives. However, jointly optimizing module sequences and parameters remains challenging. Existing approaches rely on neural architecture search (NAS) or step-wise reinforcement learning (RL), but NAS suffers from a training-inference mismatch, while step-wise RL leads to unstable training and high computational overhead due to stage-wise decision-making. We propose POS-ISP, a sequence-level RL framework that formulates modular ISP optimization as a global sequence prediction problem. Our method predicts the entire module sequence and its parameters in a single forward pass and optimizes the pipeline using a terminal task reward, eliminating the need for intermediate supervision and redundant executions. Experiments across multiple downstream tasks show that POS-ISP improves task performance while reducing computational cost, highlighting sequence-level optimization as a stable and efficient paradigm for task-aware ISP. The project page is available at https://w1jyun.github.io/POS-ISP
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{won2026pos,
title = {POS-ISP: Pipeline Optimization at the Sequence Level for Task-aware ISP},
author = {Jiyun Won and Heemin Yang and Woohyeok Kim and Jungseul Ok and Sunghyun Cho},
year = {2026},
abstract = {Recent work has explored optimizing image signal processing (ISP) pipelines for various tasks by composing predefined modules and adapting them to task-specific objectives. However, jointly optimizing module sequences and parameters remains challenging. Existing approaches rely on neural architecture search (NAS) or step-wise reinforcement learning (RL), but NAS suffers from a training-inference mismatch, while step-wise RL leads to unstable training and high computational overhead due to stage-},
url = {https://huggingface.co/papers/2604.06938},
keywords = {neural architecture search, reinforcement learning, image signal processing, module sequences, task-specific optimization, terminal task reward, forward pass, computational overhead, code available, huggingface daily},
eprint = {2604.06938},
archiveprefix = {arXiv},
}
{}