XingruiWang / 3D-Aware-VQA

Official Code for the NeurIPS'23 paper "3D-Aware Visual Question Answering about Parts, Poses and Occlusions"

3D-Aware Visual Question Answering about Parts, Poses and Occlusions (NeurIPS 2023)

Xingrui Wang, Wufei Ma, Zhuowan Li, Adam Kortylewski, Alan Yuille

Arxiv / Dataset / Bibtex

Description

This is the implementation of the NeurIPS'23 paper "3D-Aware Visual Question Answering about Parts, Poses and Occlusions". This paper contains two parts:

1. Super-CLEVR-3D. A compositional reasoning dataset that contains questions about object parts, their 3D poses, and occlusions. $\rightarrow$ ./superclevr-3D-question
2. PO3D-VQA. A 3D-aware VQA model, combining 3D generative representations of objects for robust visual recognition and probabilistic neural symbolic program execution for reasoning. $\rightarrow$ ./PO3D-VQA

Citation

@article{wang20233d,
  title={3D-Aware Visual Question Answering about Parts, Poses and Occlusions},
  author={Wang, Xingrui and Ma, Wufei and Li, Zhuowan and Kortylewski, Adam and Yuille, Alan},
  journal={arXiv preprint arXiv:2310.17914},
  year={2023}
}