Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Learning Group Interactions and Semantic Intentions for Multi-Object Trajectory Prediction
Mengshi Qi, Yuxin Yang, Huadóng Ma
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use diffusion framework instead of MLP encoder-decoder (ablation) | trajectory predictor · NBA SportVU | minADE/minFDE at 4.0s: 5.28/8.71 -> 0.84/1.17 | Improved |
| use diffusion framework instead of learnable-query attention (ablation) | trajectory predictor · NBA SportVU | minADE/minFDE at 4.0s: 1.24/1.63 -> 0.84/1.17 | Improved |
| add Banzhaf Interaction module (team-level) (ablation) | tactic prediction · NBA SportVU | Top-1 accuracy: 58.44 -> 60.09 | Improved |
| add Banzhaf Interaction module (scene-level) (ablation) | tactic prediction · NBA SportVU | Top-1 accuracy: 58.44 -> 60.23 | Improved |
| use scene-level instead of team-level Banzhaf learner (ablation) | tactic prediction · NBA SportVU | Top-1 accuracy: 60.09 -> 60.23 | Improved |
| use group tactic conditioning and Banzhaf intention vs LED | trajectory predictor · NBA SportVU | minADE at 4.0s: 0.89 -> 0.84 | Improved |
| use method vs LED on soccer | trajectory predictor · TeamTrack-Soccer | minADE/minFDE at 4.8s: 1.75/3.01 -> 1.48/2.72 | Improved |
| use method vs LED on basketball | trajectory predictor · TeamTrack-Basketball | minADE at 4.8s: 2.01 -> 1.72 | Improved |
| add one-hot tactic embedding to baseline | Transformer tactic predictor · NBA SportVU | Top-1 accuracy: 34.56 -> 53.31 | Improved |
| add team pooling to tactic-embedding baseline | Transformer tactic predictor · NBA SportVU | Top-1 accuracy: 53.31 -> 55.22 | Improved |
| LoRA finetune Llama3-8B on the benchmark | Llama3-8B · NBA SportVU | Top-1 accuracy: 22.39 -> 38.17 | Improved |
| use proposed method instead of finetuned Llama3-8B | tactic predictor · NBA SportVU | Top-1 accuracy: 38.17 -> 60.23 | Improved |
Closest papers by meaning
- Where Will Players Move Next? Dynamic Graphs and Hierarchical Fusion for Movement Forecasting in Badminton (AAAI 2023 · similarity 0.87)
- Sports-Traj: A Unified Trajectory Generation Model for Multi-Agent Movement in Sports (ICLR 2025 · similarity 0.87)
- SocialMOIF: Multi-Order Intention Fusion for Pedestrian Trajectory Prediction (CVPR 2025 · similarity 0.87)
- Event2Tracking: Reconstructing Multi-Agent Soccer Trajectories Using Long-Term Multimodal Context (AAAI 2025 · similarity 0.86)
- Higher-order Relational Reasoning for Pedestrian Trajectory Prediction (CVPR 2024 · similarity 0.85)
- Probabilistic Modeling of Interpersonal Coordination Processes (ICML 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
