Reproduce “3D-SceneQ: Empowering 3D LLM With Query-Guided Adaptive Pruning and Multi-Modal Feature Enhancement” 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
Yongqi Shan, Lu Zhang, Jiazuo Yu, Yunzhi Zhuge, Huchuan Lu
3D-SceneQ addresses 3D scene question answering by selecting question-relevant objects and enriching their 3D features with image-derived visual information. QGAP ranks objects and predicts how many to retain, while MOFE fuses image and 3D features. On SQA3D test, it reports 67.0% EM@1 versus LEO’s 50.0%; QGAP-only reaches 70.1%, so the full model is not best on this metric.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.