LMM-PCQA: Assisting Point Cloud Quality Assessment with LMM

ACM MM2024 Best Paper Nomination

Zicheng Zhang¹, Haoning Wu², Yingjie Zhou¹, Chunyi Li^1*, Wei Sun¹,

Chaofeng Chen², Xiongkuo Min^1#, Xiaohong Liu¹, Weisi Lin², Guangtao Zhai^1#

¹Shanghai Jiaotong University, ²Nanyang Technological University

^#Corresponding authors.

Paper | Github | Data

How to infer PCQA with LMM

This part only introduces the usage of the projection branch with LMM.

1. Prepare the environment of Q-Align.

2. Get the LMM weights: SJTU WPC WPC2.0

3. Download the projections and meta information from Huggingface.

4. Modify the necessary input args of the `pcqa_eval.py' file in this repo and begin the inference with Q-Align.

Contact

Please contact any of the first authors of this paper for queries.

• Zicheng Zhang, zzc1998@sjtu.edu.cn, @zzc-1998

Citation

If you find our work interesting, please feel free to cite our paper:

@article{zhang2024lmm,
  title={LMM-PCQA: Assisting Point Cloud Quality Assessment with LMM},
  author={Zhang, Zicheng and Wu, Haoning and Zhou, Yingjie and Li, Chunyi and Sun, Wei and Chen, Chaofeng and Min, Xiongkuo and Liu, Xiaohong and Lin, Weisi and Zhai, Guangtao},
  journal={ACM MM},
  year={2024}
}