Research map / Video / Video segmentation and tracking
Low-Latency Video Semantic Segmentation
Read the paperCheck 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 adaptive key-frame selection to adaptive feature propagation (ablation) | AFP + fix schedule · Cityscapes | mIOU: 75.26% -> 76.84% | Helped |
| Add low-latency scheduling (ablation) | AFP + AKS · Cityscapes | Latency: 380ms -> 119ms | Helped |
| Add low-latency scheduling (ablation) | AFP + AKS · Cityscapes | mIOU: 76.84% -> 75.89% | Hurt |
| Replace fixed-interval key-frame scheduling with adaptive key-frame selection (ablation) | Adaptive feature propagation · Cityscapes | mIOU: 75.26% -> 76.84% | Helped |
| Fuse current-frame low-level features using AdaptNet (ablation) | Adaptive Propagation Module · Cityscapes | mIOU: 68.41% -> 75.26% | Helped |
| Use spatially variant learned propagation weights instead of globally learned unified weights (ablation) | Adaptive Propagation Module · Cityscapes | mIOU: 58.73% -> 75.26% | Helped |
| Learn propagation weights instead of setting them from image differences (ablation) | Adaptive Propagation Module · Cityscapes | mIOU: 60.12% -> 75.26% | Helped |
| Use adaptive feature propagation with a fixed-interval schedule instead of per-frame full-network inference (ablation) | ResNet-101 video segmentation · Cityscapes | Avg RT: 360ms -> 151ms | Helped |
| Use adaptive feature propagation with a fixed-interval schedule instead of per-frame full-network inference (ablation) | ResNet-101 video segmentation · Cityscapes | mIOU: 80.2% -> 75.26% | Hurt |
| Choose adaptive scheduling rather than fixed-rate or threshold scheduling (ablation) | Adaptive feature propagation · Cityscapes | mIOU | Helped |
| Use the complete proposed video segmentation framework | ResNet-101 video segmentation · CamVid | Pixel Accuracy: 94.2% -> 94.6% | Helped |
| Use the complete proposed video segmentation framework | ResNet-101 video segmentation · CamVid | CA: 82.5% -> 82.9% | Helped |
Closest papers by meaning
- Semantic Video Segmentation by Gated Recurrent Flow Propagation (CVPR 2018 · similarity 0.91)
- Exploiting Temporal State Space Sharing for Video Semantic Segmentation (CVPR 2025 · similarity 0.89)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.88)
- Exploring Temporal Feature Correlation for Efficient and Stable Video Semantic Segmentation (AAAI 2024 · similarity 0.88)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 2025 · similarity 0.88)
- Efficient Event-Based Semantic Segmentation via Exploiting Frame-Event Fusion: A Hybrid Neural Network Approach (AAAI 2025 · 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.