Research map / Detection and segmentation / Weakly supervised detection and segmentation
Instance-Level Video Segmentation From Object Tracks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 semantic segmentation features to the discriminative term (ablation) | Ours · StreetDance 3D test | F1: 78.3% -> 80.1% | Helped |
| Add temporal edges to the graph (ablation) | Ours · StreetDance 3D test | F1: 76.4% -> 78.3% | Helped |
| Process all frames jointly rather than one by one (ablation) | Ours · StreetDance 3D test | F1: 76.4% -> 78.3% | Helped |
| Add the discriminative term to the grouping term (ablation) | Ours · StreetDance 3D test | F1: 77.6% -> 78.3% | Helped |
| Add the grouping term to the discriminative term (ablation) | Ours · StreetDance 3D test | F1: 66.9% -> 78.3% | Helped |
| Add segmentation constraints based on object tracks (ablation) | Ours · StreetDance 3D test | F1: 12.8% -> 78.3% | Helped |
| Add non-convex refinement after convex optimization (ablation) | Ours · StreetDance 3D test | F1: 75.6% -> 78.3% | Helped |
| Add disparity features (ablation) | Ours · StreetDance 3D test | F1: 74.0% -> 78.3% | Helped |
| Use the proposed method rather than CRF as RNN | Instance-level segmentation with automatic tracks · StreetDance 3D test, automatic tracks | F1: 56.2% -> 63.6% | Helped |
| Use the proposed method rather than SDS | Instance-level segmentation with SDS detections · StreetDance 3D test, SDS detections | F1: 65.1% -> 72.5% | Helped |
| Use the proposed method rather than the state-of-the-art method [48] | Drifting car video segmentation · Drifting car | Overlap: 79.9% -> 86.8% | Helped |
Closest papers by meaning
- Weakly Supervised Instance Segmentation for Videos With Temporal Mask Consistency (CVPR 2021 · similarity 0.90)
- Self-supervised Object-Centric Learning for Videos (NeurIPS 2023 · similarity 0.88)
- Video Segmentation via Multiple Granularity Analysis (CVPR 2017 · similarity 0.88)
- Minimizing Labeled, Maximizing Unlabeled: An Image-Driven Approach for Video Instance Segmentation (CVPR 2025 · similarity 0.88)
- Learning to Track Instances without Video Annotations (CVPR 2021 · similarity 0.88)
- Video Object Segmentation Using Global and Instance Embedding Learning (CVPR 2021 · similarity 0.87)
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.