Research map / Label-efficient and robust learning / Recognition and long-tail
TF-FACE: Time-Frequency Fusion Learning via Frequency-Domain Adaptive and Controllable Enhancement for Trajectory Prediction
Dongjian Song, Yunhao Meng, Songjun Huang, Jiayi Han
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 |
|---|---|---|---|
| add gated frequency-domain attention (GfreAttn) encoder (ablation) | TF-FACE · Argoverse 1 val | Brier minFDE6 | Improved |
| add low-frequency global feature extraction (LFE, freMLP) (ablation) | TF-FACE · Argoverse 1 val | minFDE6 | Improved |
| add high-frequency local feature extraction (HFE) branch (ablation) | TF-FACE · Argoverse 1 val | minADE6 | Improved |
| add dual-branch parallel decoder (DuaD) stage (ablation) | TF-FACE · Argoverse 1 val | Brier minFDE6 | Improved |
| use full TF-FACE with all four modules (ablation) | TF-FACE · Argoverse 1 val | Brier minFDE6: 1.496 (full) vs 1.549 to 1.552 ablated | Improved |
| add random-seed ensemble | TF-FACE · Argoverse 1 test | Brier minFDE6: 1.73 -> 1.676 | Improved |
| use TF-FACE instead of FIM | FIM · Argoverse 2 val (30 frames) | Brier minFDE6: 1.131 -> 1.092 | Improved |
| use TF-FACE instead of QCNet on long horizon | QCNet · Argoverse 2 val (60 frames) | minFDE6: 1.29 -> 1.31 | No clear effect |
| use TF-FACE instead of QCNet on long horizon | QCNet · Argoverse 2 val (60 frames) | MR6: 0.167 -> 0.164 | Improved |
| use TF-FACE instead of HPNet | HPNet · Argoverse 1 test | DAC: 0.988 -> 0.9911 | Improved |
Closest papers by meaning
- C2F-TP: A Coarse-to-Fine Denoising Framework for Uncertainty-Aware Trajectory Prediction (AAAI 2025 · similarity 0.84)
- ModeSeq: Taming Sparse Multimodal Motion Prediction with Sequential Mode Modeling (CVPR 2025 · similarity 0.84)
- DeMo: Decoupling Motion Forecasting into Directional Intentions and Dynamic States (NeurIPS 2024 · similarity 0.84)
- T4P: Test-Time Training of Trajectory Prediction via Masked Autoencoder and Actor-specific Token Memory (CVPR 2024 · similarity 0.83)
- Trajectory Mamba: Efficient Attention-Mamba Forecasting Model Based on Selective SSM (CVPR 2025 · similarity 0.83)
- UniMotion: A Unified Motion Framework for Simulation, Prediction and Planning (NeurIPS 2025 · similarity 0.83)
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.
