learning-4d-reasoning

【数据与研究】Enable VLMs to perform dynamic spatial reasoning (DSR) by extracting 4D priors from videos and using Geometry Selection Modules (GSM) for selective injection. Provides DSR-Train dataset (50K QA pairs) and benchmark with six reasoning types, balancing geometric specialization with general video understanding—improving VLM 4D reasoning without degradat…

独立 GitHub 快照 · 自动分类 · 以原作者文档为准