Research map / Video / Video segmentation and tracking
Video Propagation Networks
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 a second VPN refinement stage (ablation) | VPN for DAVIS video object segmentation · DAVIS | IoU: 70.1 -> 71.3 | Helped |
| Add a second VPN refinement stage (ablation) | VPN for DAVIS video object segmentation · DAVIS | contour accuracy (F): 68.4 -> 68.9 | Helped |
| Train additional stages beyond VPN-Stage2 (ablation) | VPN for DAVIS video object segmentation · DAVIS | IoU | No clear effect |
| Use learned bilateral filters instead of identity filters (ablation) | VPN for DAVIS video object segmentation · DAVIS | IoU: 67.0 -> 70.1 | Helped |
| Use superpixels rather than random point sampling (ablation) | BNN-Identity · DAVIS video object segmentation | IoU: -0.5 | Hurt |
| Use superpixel sampling to reduce input points (ablation) | BNN-Identity · DAVIS video object segmentation | number of input points: a factor of 10 | Helped |
| Use optical-flow-warped position features (ablation) | VPN with CNN-1 · CamVid test | IoU: 66.5 -> 66.7 | Helped |
| Add trained VPN rather than BNN-Identity (ablation) | CNN-1 · CamVid test | IoU: 65.3 -> 66.5 | Helped |
| Add VPN-Flow to CNN-2 (ablation) | CNN-2 · CamVid test | IoU: 68.9 -> 69.5 | Helped |
| Replace bilateral filters with spatial filters (ablation) | VPN with CNN-1 · CamVid test | IoU: 0.3 IoU | Hurt |
| Use learned bilateral filters instead of identity filters (ablation) | VPN for video color propagation · DAVIS video color propagation | PSNR: 27.89 -> 28.15 | Helped |
| Train more than one VPN stage (ablation) | VPN for video color propagation · DAVIS video color propagation | PSNR | No clear effect |
Closest papers by meaning
- Semantic Video Segmentation by Gated Recurrent Flow Propagation (CVPR 2018 · similarity 0.87)
- Improving Semantic Segmentation via Video Propagation and Label Relaxation (CVPR 2019 · similarity 0.86)
- Low-Latency Video Semantic Segmentation (CVPR 2018 · similarity 0.86)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.85)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.84)
- Pay Attention to Target: Relation-Aware Temporal Consistency for Domain Adaptive Video Semantic Segmentation (AAAI 2024 · similarity 0.84)
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.