Research map / Video / Video segmentation and tracking
Knowledge Transfer for Scene-Specific Motion Prediction
Lamberto Ballan, Francesco Castaldo, Alexandre Alahi, Francesco Palmieri, Silvio Savarese
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 |
|---|---|---|---|
| train the model on the same scene instead of transferring knowledge from other scenes | Ours (DBN) · Stanford-UAV | MHD error: 14.29 -> 8.44 | Improved |
| use the scene-aware DBN instead of linear constant-velocity prediction | Ours (DBN) · UCLA-courtyard | MHD error: 41.36 -> 10.32 | Improved |
| use the scene-aware DBN instead of linear constant-velocity prediction | Ours (DBN) · Stanford-UAV | MHD error: 31.29 -> 8.44 | Improved |
| add collision avoidance to the linear prediction baseline | LP · Stanford-UAV | MHD error: 31.29 -> 21.30 | Improved |
| replace patch matching with learned context descriptors for knowledge transfer (ablation) | Ours (transfer) · Stanford-UAV | MHD error: 21.67 -> 14.29 | Improved |
| transfer gain on pedestrians versus cyclists compared with linear prediction | Ours (transfer) · Stanford-UAV | MHD error: 34.48 -> 12.36 (pedestrian), 28.09 -> 16.22 (cyclist) | Mixed |
| use ground-truth segmentation masks instead of scene parsing output (ablation) | Ours (transfer) · Stanford-UAV | MHD error | No clear effect |
Closest papers by meaning
- Predicting Scene Parsing and Motion Dynamics in the Future (NeurIPS 2017 · similarity 0.90)
- An Uncertain Future: Forecasting from Static Images Using Variational Autoencoders (ECCV 2016 · similarity 0.89)
- Schroedinger’s Cat: Probabilistic Representation and Prediction of Potential Scene Kinematics (ECCV 2026 · similarity 0.89)
- Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs (NeurIPS 2026 · similarity 0.87)
- PlaySlot: Learning Inverse Latent Dynamics for Controllable Object-Centric Video Prediction and Planning (ICML 2025 · similarity 0.87)
- E-Motion: Future Motion Simulation via Event Sequence Diffusion (NeurIPS 2024 · similarity 0.86)
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.
