Reproduce “Flow Matching Reinforcement for 3D Mesh Generation via Dynamic Homing Optimization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhen Zhou, Zhiwei Ning, Puhua Jiang, Sheng Zhang, Yifei Tang, Jie Yang, Xintong Han, Wei Liu and 1 more
Flow3D-Pro applies Dynamic Homing Optimization (DHO) to image-to-3D mesh generation, which pairs each dispreferred sample with a distinct preferred sample and steers the former toward its assigned target. In comparison on 200 images, its Stage II-refined outputs scored 4.26 in expert rankings, versus 3.00 for the strongest competing reinforcement-learning method. This relative preference result is limited to that evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.