VEvo-Harness improves self-updating AI for precise image creation
VEvo-Harness: Causally-Guarded Self-Evolving Harness for Agentic Image Generation
Computer Vision and Pattern Recognition
Summary
Creating images that perfectly match detailed user requests is difficult because current AI systems don't learn well from their mistakes and repairs. This paper introduces VEvo-Harness, a system that helps AI image generators learn continuously and improve over time by carefully tracking mistakes, verifying fixes, and storing useful skills. It uses a smart process to find the real cause of problems, tests possible fixes before applying them, and chooses the best way to repair images. The authors show that this approach works better than existing AI tools and matches some commercial image generators.
What this means in practice
- •For creative design teams: Enable design tools to automatically improve image edits by learning from past failures and fixes, reducing manual corrections.
- •For automated content generation services: Support continual enhancement of AI image generation quality by integrating causally verified repair skills for user-specified constraints.$Commercial implications: It makes selling advanced image creation services with self-improving AI possible by ensuring reliable adaptation to user demands.
Authors
Yuan Wang, Yuhang Yu, Mengting Chen, Tianrun Xu, Zihao Pan, Xuetao Feng, Wei Li
Abstract
The central bottleneck in visual generation and editing has migrated from achieving pixel-level fidelity to satisfying densely constrained user specifications through increasingly agentic workflows. Prevailing agentic generation pipelines remain shallow generate-verify-rewrite loops treating VLM judgments as coarse visual feedback, either discarding repair experience or admitting ungated updates that poison skill library. With memorize and govern periods left open, they degrade into ungoverned iterators of repeated trial-and-error rather than continual self-evolving harness. Towards this issue, we present \textbf{VEvo-Harness}, the first causally-guarded continual self-evolving harness system for agentic visual generation. VEvo-Harness introduces four coupled innovations: (1) \textit{a visual traceback} that densifies sparse judgments into root-cause attributions with a hierarchical verification tree; (2) \textit{a visual continual governance} that admits only causally valid skill patches through counterfactual dual-arm image generation replay; (3) \textit{a lifecycle-managed harness state} that hierarchically distills root-cause-anchored episodes across working, experience, and pattern tiers into a persistent skill library of routable and causally verified capabilities; (4) \textit{a heterogeneous sandbox policy} that orchestrates a diverse action space of repair primitives, dynamically routing cost-aware interventions based on root causes to select the most effective strategy. VEvo-Harness surpasses state-of-the-art agentic methods and rivals proprietary generators (\textit{e.g.}, GPT-Image, Nano-Pro) while maintaining the continual harness evolution and generalization.