Research map / Video / Video segmentation and tracking
Long-Term Image Boundary Prediction
Apratim Bhattacharyya, Mateusz Malinowski, Bernt Schiele, Mario Fritz
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 context to the multi-scale predictor (ablation) | CMSC · VSB100, long-term prediction | boundary F-measure | Improved |
| use multiple scales instead of a single scale (ablation) | CMSC · VSB100 | boundary F-measure | Improved |
| add table-border information rather than hide table borders (ablation) | CMSC · Single-ball billiard worlds, t + 20 | best F-measure: 0.347 -> 0.632 | Improved |
| train for the test-world ball count rather than use the mismatched training count (ablation) | CMSC · Six-ball worlds, t + 20 | best F-measure: 0.511 -> 0.600 | Improved |
| apply a ball-tracker mask when evaluating real billiard motion (ablation) | CMSC · Real billiard sequences, t + 20 | F-measure: 0.087 -> 0.497 | Improved |
| fuse predicted boundaries with predicted RGB frames | RGB prediction · VSB100, t + 4 | PSNR: 20.4 -> 22.3 | Improved |
| fuse predicted boundaries with predicted RGB frames | RGB prediction · UCF101, t + 4 | PSNR: 21.4 -> 23.8 | Improved |
| use CMSC instead of the constant last-frame prediction | Single-ball boundary prediction · Single-ball billiard worlds, t + 20 | best F-measure: 0.002 -> 0.632 | Improved |
| use CMSC instead of the constant last-frame prediction | Two-ball boundary prediction · Two-ball worlds, t + 20 | best F-measure: 0.101 -> 0.681 | Improved |
| use CMSC instead of the constant last-frame prediction | Three-ball boundary prediction · Three-ball worlds, t + 20 | best F-measure: 0.090 -> 0.700 | Improved |
| use CMSC instead of the constant last-frame prediction | Real billiard boundary prediction · Real billiard sequences, t + 20 | F-measure: 0.844 -> 0.746 | Hurt |
Closest papers by meaning
- Boundary Flow: A Siamese Network That Predicts Boundary Motion Without Training on Motion (CVPR 2018 · similarity 0.87)
- Temporal Perceiver: A General Architecture for Arbitrary Boundary Detection (TPAMI 2023 · similarity 0.87)
- Disentangling Propagation and Generation for Video Prediction (ICCV 2019 · similarity 0.87)
- Generative Perception of Shape and Material from Differential Motion (NeurIPS 2025 · similarity 0.86)
- Generative Image Dynamics (CVPR 2024 · similarity 0.85)
- VideoJAM: Joint Appearance-Motion Representations for Enhanced Motion Generation in Video Models (ICML 2025 · 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.
