Reproduce “Hyper-STTN: Hypergraph Augmented Spatial-Temporal Transformer Network for Trajectory Prediction” 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

Weizheng Wang, Baijian Yang, Sungeun Hong, Wenhai Sun, Byung-Cheol Min

The authors introduce Hyper-STTN to forecast crowd trajectories by combining pairwise spatial and temporal transformers with multiscale hypergraphs that represent group interactions. Cross-modal attention aligns the representations before a conditional variational autoencoder generates predictions. On ETH-UCY’s five subsets, its aggregate minADE20/minFDE20 scores are 0.21/0.33, below STTN’s 0.31/0.43 and HGNN’s 0.29/0.48; results do not establish superiority on every subset.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.