Research map / Video / Video segmentation and tracking
Coarse-To-Fine Feature Mining for Video Semantic Segmentation
Guolei Sun, Yun Liu, Henghui Ding, Thomas Probst, Luc Van Gool
Check 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 CFFM to SegFormer | SegFormer with MiT-B0 · VSPW validation | mIoU: 32.9 -> 35.4 | Improved |
| Add CFFM to SegFormer | SegFormer with MiT-B1 · VSPW validation | mIoU: 36.5 -> 38.5 | Improved |
| Add CFFM to SegFormer | SegFormer with MiT-B2 · VSPW validation | mIoU: 43.9 -> 44.9 | Improved |
| Add CFFM to SegFormer | SegFormer with MiT-B5 · VSPW validation | mIoU: 48.2 -> 49.3 | Improved |
| Add CFFM to SegFormer | SegFormer with MiT-B0 · Cityscapes | mIoU: 71.9 -> 74.0 | Improved |
| Add CFFM to SegFormer | SegFormer with MiT-B1 · Cityscapes | mIoU: 74.1 -> 75.1 | Improved |
| Add CFFM to SegFormer | SegFormer with MiT-B0 · VSPW validation | mVC16: 77.3 -> 82.9 | Improved |
| Increase cross-frame attention layers from 1 to 2 (ablation) | CFFM with MiT-B1 · VSPW validation | mIoU: 37.8 -> 38.5 | Improved |
| Increase cross-frame attention layers from 2 to 3 (ablation) | CFFM with MiT-B1 · VSPW validation | mIoU: 38.5 -> 38.7 | Improved |
| Increase cross-frame attention layers from 3 to 4 (ablation) | CFFM with MiT-B1 · VSPW validation | mIoU: 38.7 -> 38.8 | Improved |
| Use one more distant reference frame, moving k3 from 3 to 6 (ablation) | CFFM with MiT-B1 · VSPW validation | mIoU: 37.4 -> 37.7 | Improved |
| Move the single reference frame from k3=6 to k3=9 (ablation) | CFFM with MiT-B1 · VSPW validation | mIoU: 37.7 -> 37.9 | Improved |
Closest papers by meaning
- Learning Local and Global Temporal Contexts for Video Semantic Segmentation (TPAMI 2024 · similarity 0.95)
- Exploring Temporal Feature Correlation for Efficient and Stable Video Semantic Segmentation (AAAI 2024 · similarity 0.88)
- Mining Relations among Cross-Frame Affinities for Video Semantic Segmentation (ECCV 2022 · similarity 0.88)
- MCIBI++: Soft Mining Contextual Information Beyond Image for Semantic Segmentation (TPAMI 2022 · similarity 0.87)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.86)
- Learning Spatial-Semantic Features for Robust Video Object Segmentation (ICLR 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.
