Mesh-RFT: Enhancing Mesh Generation via Fine-grained Reinforcement Fine-Tuning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Liu, Jian, Xu, Jing, Guo, Song, Li, Jing, Guo, Jingfeng, Yu, Jiaao, Weng, Haohan, Lei, Biwen, Yang, Xianghui, Chen, Zhuo, Zhu, Fangqi, Han, Tao, Guo, Chunchao
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917034858643456
author Liu, Jian
Xu, Jing
Guo, Song
Li, Jing
Guo, Jingfeng
Yu, Jiaao
Weng, Haohan
Lei, Biwen
Yang, Xianghui
Chen, Zhuo
Zhu, Fangqi
Han, Tao
Guo, Chunchao
author_facet Liu, Jian
Xu, Jing
Guo, Song
Li, Jing
Guo, Jingfeng
Yu, Jiaao
Weng, Haohan
Lei, Biwen
Yang, Xianghui
Chen, Zhuo
Zhu, Fangqi
Han, Tao
Guo, Chunchao
contents Existing pretrained models for 3D mesh generation often suffer from data biases and produce low-quality results, while global reinforcement learning (RL) methods rely on object-level rewards that struggle to capture local structure details. To address these challenges, we present Mesh-RFT, a novel fine-grained reinforcement fine-tuning framework that employs Masked Direct Preference Optimization (M-DPO) to enable localized refinement via quality-aware face masking. To facilitate efficient quality evaluation, we introduce an objective topology-aware scoring system to evaluate geometric integrity and topological regularity at both object and face levels through two metrics: Boundary Edge Ratio (BER) and Topology Score (TS). By integrating these metrics into a fine-grained RL strategy, Mesh-RFT becomes the first method to optimize mesh quality at the granularity of individual faces, resolving localized errors while preserving global coherence. Experiment results show that our M-DPO approach reduces Hausdorff Distance (HD) by 24.6% and improves Topology Score (TS) by 3.8% over pre-trained models, while outperforming global DPO methods with a 17.4% HD reduction and 4.9% TS gain. These results demonstrate Mesh-RFT's ability to improve geometric integrity and topological regularity, achieving new state-of-the-art performance in production-ready mesh generation. Project Page: https://hitcslj.github.io/mesh-rft/.
format Preprint
id arxiv_https___arxiv_org_abs_2505_16761
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mesh-RFT: Enhancing Mesh Generation via Fine-grained Reinforcement Fine-Tuning
Liu, Jian
Xu, Jing
Guo, Song
Li, Jing
Guo, Jingfeng
Yu, Jiaao
Weng, Haohan
Lei, Biwen
Yang, Xianghui
Chen, Zhuo
Zhu, Fangqi
Han, Tao
Guo, Chunchao
Computer Vision and Pattern Recognition
Existing pretrained models for 3D mesh generation often suffer from data biases and produce low-quality results, while global reinforcement learning (RL) methods rely on object-level rewards that struggle to capture local structure details. To address these challenges, we present Mesh-RFT, a novel fine-grained reinforcement fine-tuning framework that employs Masked Direct Preference Optimization (M-DPO) to enable localized refinement via quality-aware face masking. To facilitate efficient quality evaluation, we introduce an objective topology-aware scoring system to evaluate geometric integrity and topological regularity at both object and face levels through two metrics: Boundary Edge Ratio (BER) and Topology Score (TS). By integrating these metrics into a fine-grained RL strategy, Mesh-RFT becomes the first method to optimize mesh quality at the granularity of individual faces, resolving localized errors while preserving global coherence. Experiment results show that our M-DPO approach reduces Hausdorff Distance (HD) by 24.6% and improves Topology Score (TS) by 3.8% over pre-trained models, while outperforming global DPO methods with a 17.4% HD reduction and 4.9% TS gain. These results demonstrate Mesh-RFT's ability to improve geometric integrity and topological regularity, achieving new state-of-the-art performance in production-ready mesh generation. Project Page: https://hitcslj.github.io/mesh-rft/.
title Mesh-RFT: Enhancing Mesh Generation via Fine-grained Reinforcement Fine-Tuning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.16761