Research map / 3D vision / Human pose estimation
Instance-Level Future Motion Estimation in a Single Image Based on Ordinal Regression
Kyung-Rae Kim, Whan Choi, Yeong Jun Koh, Seong-Gyun Jeong, Chang-Su Kim
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 global context feature to the object feature (MCP layer) for direction (ablation) | FM network · FM dataset (pedestrians) | direction accuracy (%): 75.9 -> 76.8 | Improved |
| add object feature to global-only network for direction (ablation) | FM network · FM dataset (pedestrians) | direction accuracy (%): 25.8 -> 76.8 | Improved |
| add global feature to object feature for speed (ablation) | FM network · FM dataset (pedestrians) | speed accuracy (%): 88.3 -> 90.1 | Improved |
| add global feature to object feature for action (ablation) | FM network · FM dataset (pedestrians) | action accuracy (%): 86.8 -> 86.7 | No clear effect |
| replace multi-class direction classification with cyclic ordinal regression (ablation) | Object + Global network · FM dataset (pedestrians) | direction accuracy (%): 76.8 -> 77.7 | Improved |
| replace multi-class direction classification with cyclic ordinal regression (ablation) | Object + Global network · FM dataset (pedestrians) | Direction+ accuracy (%): 96.0 -> 96.6 | Improved |
| replace multi-class speed classification with linear ordinal regression (ablation) | Object + Global network · FM dataset (pedestrians) | speed accuracy (%): 90.1 -> 90.4 | Improved |
| train with ordinal-regression direction and speed heads (action head unchanged) (ablation) | Object + Global network · FM dataset (pedestrians) | action accuracy (%): 86.7 -> 87.2 | Improved |
| narrow MDNet search region with predicted future motion (FM) | MDNet · TC128+OTB pedestrian sequences (setting IV) | fps: 1.23 -> 2.15 | Improved |
| narrow MDNet search region with predicted future motion (FM) | MDNet · TC128+OTB pedestrian sequences (setting IV) | precision (PR): 0.875 -> 0.893 | Improved |
| narrow MDNet search region with predicted future motion (FM) | MDNet · TC128+OTB pedestrian sequences (setting III) | precision (PR): 0.883 -> 0.830 | Hurt |
| narrow CDT search region with predicted future motion plus background compensation | CDT · MOT17 (5 fps and 1 fps) | MOTA | Improved |
Closest papers by meaning
- An Uncertain Future: Forecasting from Static Images Using Variational Autoencoders (ECCV 2016 · similarity 0.86)
- Object Tracking using Spatio-Temporal Networks for Future Prediction Location (ECCV 2020 · similarity 0.86)
- Probabilistic Future Prediction for Video Scene Understanding (ECCV 2020 · similarity 0.85)
- Jointly learning visual motion and confidence from local patches in event cameras (ECCV 2020 · similarity 0.85)
- SmartPretrain: Model-Agnostic and Dataset-Agnostic Representation Learning for Motion Prediction (ICLR 2025 · similarity 0.83)
- E-Motion: Future Motion Simulation via Event Sequence Diffusion (NeurIPS 2024 · 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.
