Evolving, Not Training: Zero-Shot Reasoning Segmentation via Evolutionary Prompting

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Ye, Kai, You, Xiaotong, Lin, Jianghang, Ji, Jiayi, Dai, Pingyang, Cao, Liujuan
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911347731595264
author Ye, Kai
You, Xiaotong
Lin, Jianghang
Ji, Jiayi
Dai, Pingyang
Cao, Liujuan
author_facet Ye, Kai
You, Xiaotong
Lin, Jianghang
Ji, Jiayi
Dai, Pingyang
Cao, Liujuan
contents Reasoning Segmentation requires models to interpret complex, context-dependent linguistic queries to achieve pixel-level localization. Current dominant approaches rely heavily on Supervised Fine-Tuning (SFT) or Reinforcement Learning (RL). However, SFT suffers from catastrophic forgetting and domain dependency, while RL is often hindered by training instability and rigid reliance on predefined reward functions. Although recent training-free methods circumvent these training burdens, they are fundamentally limited by a static inference paradigm. These methods typically rely on a single-pass "generate-then-segment" chain, which suffers from insufficient reasoning depth and lacks the capability to self-correct linguistic hallucinations or spatial misinterpretations. In this paper, we challenge these limitations and propose EVOL-SAM3, a novel zero-shot framework that reformulates reasoning segmentation as an inference-time evolutionary search process. Instead of relying on a fixed prompt, EVOL-SAM3 maintains a population of prompt hypotheses and iteratively refines them through a "Generate-Evaluate-Evolve" loop. We introduce a Visual Arena to assess prompt fitness via reference-free pairwise tournaments, and a Semantic Mutation operator to inject diversity and correct semantic errors. Furthermore, a Heterogeneous Arena module integrates geometric priors with semantic reasoning to ensure robust final selection. Extensive experiments demonstrate that EVOL-SAM3 not only substantially outperforms static baselines but also significantly surpasses fully supervised state-of-the-art methods on the challenging ReasonSeg benchmark in a zero-shot setting. The code is available at https://github.com/AHideoKuzeA/Evol-SAM3.
format Preprint
id arxiv_https___arxiv_org_abs_2512_24702
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Evolving, Not Training: Zero-Shot Reasoning Segmentation via Evolutionary Prompting
Ye, Kai
You, Xiaotong
Lin, Jianghang
Ji, Jiayi
Dai, Pingyang
Cao, Liujuan
Computer Vision and Pattern Recognition
Artificial Intelligence
Reasoning Segmentation requires models to interpret complex, context-dependent linguistic queries to achieve pixel-level localization. Current dominant approaches rely heavily on Supervised Fine-Tuning (SFT) or Reinforcement Learning (RL). However, SFT suffers from catastrophic forgetting and domain dependency, while RL is often hindered by training instability and rigid reliance on predefined reward functions. Although recent training-free methods circumvent these training burdens, they are fundamentally limited by a static inference paradigm. These methods typically rely on a single-pass "generate-then-segment" chain, which suffers from insufficient reasoning depth and lacks the capability to self-correct linguistic hallucinations or spatial misinterpretations. In this paper, we challenge these limitations and propose EVOL-SAM3, a novel zero-shot framework that reformulates reasoning segmentation as an inference-time evolutionary search process. Instead of relying on a fixed prompt, EVOL-SAM3 maintains a population of prompt hypotheses and iteratively refines them through a "Generate-Evaluate-Evolve" loop. We introduce a Visual Arena to assess prompt fitness via reference-free pairwise tournaments, and a Semantic Mutation operator to inject diversity and correct semantic errors. Furthermore, a Heterogeneous Arena module integrates geometric priors with semantic reasoning to ensure robust final selection. Extensive experiments demonstrate that EVOL-SAM3 not only substantially outperforms static baselines but also significantly surpasses fully supervised state-of-the-art methods on the challenging ReasonSeg benchmark in a zero-shot setting. The code is available at https://github.com/AHideoKuzeA/Evol-SAM3.
title Evolving, Not Training: Zero-Shot Reasoning Segmentation via Evolutionary Prompting
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2512.24702