Research map / Video / Video segmentation and tracking
Video Matting via Consistency-Regularized Graph Neural Networks
Tiantian Wang, Sifei Liu, Yapeng Tian, Kai Li, Ming-Hsuan Yang
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 GNN inter-frame aggregation to image-level baseline (ablation) | CRGNN video matting · Real human matting dataset | MSE: 10.21 -> 9.480 | Improved |
| add GNN inter-frame aggregation to image-level baseline (ablation) | CRGNN video matting · Real human matting dataset | MESSDdt: 15.31 -> 13.45 | Improved |
| add consistency regularization (ablation) | CRGNN video matting · Real human matting dataset | MSE: 9.480 -> 9.260 | Improved |
| add consistency regularization (ablation) | CRGNN video matting · Real human matting dataset | SAD: 78.38 -> 73.21 | Improved |
| add consistency regularization (ablation) | CRGNN video matting · Real human matting dataset | MESSDdt: 13.45 -> 12.69 | Improved |
| add adversarial discriminator (ablation) | CRGNN video matting · Real human matting dataset | MSE: 9.260 -> 9.223 | Improved |
| add adversarial discriminator (ablation) | CRGNN video matting · Real human matting dataset | Connectivity: 60.75 -> 58.49 | Improved |
| add adversarial discriminator (ablation) | CRGNN video matting · Real human matting dataset | SAD: 73.21 -> 73.49 | No clear effect |
| use deformable alignment instead of non-local aggregation (ablation) | CRGNN video matting · Real human matting dataset | MSE: 9.954 -> 9.224 | Improved |
| use deformable alignment instead of non-local aggregation (ablation) | CRGNN video matting · Real human matting dataset | SAD: 89.45 -> 73.50 | Improved |
| increase test-time nodes from 3 to 5 (ablation) | CRGNN video matting · Real human matting dataset | MSE: 9.223 -> 9.230 | No clear effect |
| increase test-time nodes from 3 to 7 (ablation) | CRGNN video matting · Real human matting dataset | MSE: 9.223 -> 9.228 | No clear effect |
Closest papers by meaning
- Deep Video Matting via Spatio-Temporal Alignment and Aggregation (CVPR 2021 · similarity 0.91)
- MatAnyone: Stable Video Matting with Consistent Memory Propagation (CVPR 2025 · similarity 0.90)
- Adaptive Human Matting for Dynamic Videos (CVPR 2023 · similarity 0.89)
- Training Matting Models Without Alpha Labels (AAAI 2025 · similarity 0.87)
- Revisiting Context Aggregation for Image Matting (ICML 2024 · similarity 0.87)
- MaGGIe: Masked Guided Gradual Human Instance Matting (CVPR 2024 · 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.
