Research map / Detection and segmentation / Tracking
Multi-Camera Trajectory Forecasting with Trajectory Tensors
Olly Styles, Tanaya Guha, Víctor Sánchez
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 trajectory tensor input instead of coordinate trajectories (3D-CNN vs GRU) | MCTF models · WNMF (which) | AP_which: 84.0 -> 87.5 | Improved |
| add trajectory tensor input instead of coordinate trajectories (3D-CNN vs GRU) | MCTF models · WNMF (where) | AP_where: 16.5 -> 37.4 | Improved |
| add multi-view trajectory input to 3D-CNN (ablation) | 3D-CNN trajectory tensor · WNMF (which) | AP_which: 83.0 -> 87.5 | Improved |
| add multi-view trajectory input to 2D-1D-CNN (ablation) | 2D-1D-CNN trajectory tensor · WNMF (which) | AP_which: 81.8 -> 87.1 | Improved |
| add multi-view trajectory input to 3D-CNN (ablation) | 3D-CNN trajectory tensor · WNMF (when) | AP_when: 77.1 -> 79.0 | Improved |
| add multi-view trajectory input to 3D-CNN (ablation) | 3D-CNN trajectory tensor · WNMF (where) | AP_where: 38.8 -> 37.4 | Hurt |
| add multi-view trajectory input to 2D-1D-CNN (ablation) | 2D-1D-CNN trajectory tensor · WNMF (where) | AP_where: 37.4 -> 34.5 | Hurt |
| add multi-view trajectory input to CNN-GRU (ablation) | CNN-GRU trajectory tensor · WNMF (which) | AP_which: 83.4 -> 86.1 | Improved |
| use CNN-GRU instead of 3D-CNN for trajectory tensors | trajectory tensor models · WNMF (when) | AP_when: 79.0 -> 69.1 | Hurt |
| increase heatmap size with more Gaussian smoothing (ablation) | trajectory tensor models · WNMF | AP | Mixed |
| use hand-crafted features instead of learned models for displacement error | Where task · WNMF (where) | ADE_where: 216.2 vs 223.3 | Mixed |
| use 3D-CNN trajectory tensor in multi-target setting vs coordinate LSTM | multi-target MCTF · WNMF multi-target subset (which) | AP_which: 77.8 -> 80.0 | Improved |
Closest papers by meaning
- S2F2: Single-Stage Flow Forecasting for Future Multiple Trajectories Prediction (ECCV 2022 · similarity 0.86)
- Hierarchical Multi-Supervision Multi-Interaction Graph Attention Network for Multi-Camera Pedestrian Trajectory Prediction (AAAI 2022 · similarity 0.86)
- Object Tracking using Spatio-Temporal Networks for Future Prediction Location (ECCV 2020 · similarity 0.85)
- GTR: A General, Multi-View, and Dynamic Framework for Trajectory Representation Learning (ICML 2025 · similarity 0.85)
- GigaTraj: Predicting Long-term Trajectories of Hundreds of Pedestrians in Gigapixel Complex Scenes (CVPR 2024 · similarity 0.85)
- MESSENGER: Memory-Enhanced Sequential Scene Flow Estimation via Autoregressive Next-Frame Forecasting (NeurIPS 2026 · similarity 0.84)
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.
