[IROS 2025] Mesh-Learner: Texturing Mesh with Spherical Harmonics

1. Introduction

Mesh-Learner is a framework natively compatible with traditional rasterization pipelines that integrates mesh and spherical harmonic (SH) Texture (i.e., texture filled with SH coefficients) into the learning process to learn each mesh’s view-dependent radiance end-to-end.

2. Environment Setup

1 min setup: we offer the docker image ready for compilation and run

  docker pull alexwanwan/mesh_learner
  sudo docker run --net=host --ipc=host -p 10301:10400 -it -v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=:0 -e GDK_SCALE -e GDK_DPI_SCALE --name mesh_learner_docker --runtime=nvidia -v [your_path] --gpus all mesh_learner:1.0 /bin/bash
  docker exec -it mesh_learner_docker bash

3. build

  cd build && cmake .. && make -j

4. dataset

https://figshare.com/s/3a1898c06b8c99b88cc8

5. Acknowledgement

This project is relied on the following open-source projects:

• cuCollections
• rapidJson
• ReplicaSDK
• tensorboard.cpp

6. Citation

@article{wan2025mesh,
  title={Mesh-Learner: Texturing Mesh with Spherical Harmonics},
  author={Wan, Yunfei and Liu, Jianheng and Zheng, Chunran and Lin, Jiarong and Zhang, Fu},
  journal={arXiv preprint arXiv:2504.19938},
  year={2025}
}