Research map / Video / Video segmentation and tracking
Accel: A Corrective Fusion Network for Efficient Semantic Segmentation on Video
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 the ResNet-101 reference branch and warped keyframe features to DeepLab-18 (ablation) | Accel-18 · Cityscapes, keyframe interval 5, max offset | mIoU: 57.7 -> 69.2 | Helped |
| Add the update branch to the reference-only model (ablation) | Accel-18 · Cityscapes, keyframe interval 5, max offset | mIoU: 62.4 -> 69.2 | Helped |
| Add the ResNet-101 reference branch and warped keyframe features to DeepLab-34 (ablation) | Accel-34 · Cityscapes, keyframe interval 5, max offset | mIoU: 62.8 -> 69.7 | Helped |
| Add the update branch to the reference-only model (ablation) | Accel-34 · Cityscapes, keyframe interval 5, max offset | mIoU: 62.4 -> 69.7 | Helped |
| Add the ResNet-101 reference branch and warped keyframe features to DeepLab-50 (ablation) | Accel-50 · Cityscapes, keyframe interval 5, max offset | mIoU: 70.1 -> 73.0 | Helped |
| Add the update branch to the reference-only model (ablation) | Accel-50 · Cityscapes, keyframe interval 5, max offset | mIoU: 62.4 -> 73.0 | Helped |
| Add the reference branch to DeepLab-101 (ablation) | Accel-101 · Cityscapes, keyframe interval 5, max offset | mIoU: 75.2 -> 75.5 | Helped |
| Add the update branch to the reference-only model (ablation) | Accel-101 · Cityscapes, keyframe interval 5, max offset | mIoU: 62.4 -> 75.5 | Helped |
| Use feature fusion instead of score fusion (ablation) | Accel-18 · Cityscapes, keyframe interval 5, max offset | mIoU: 69.2 -> 69.5 | Helped |
| Use score fusion instead of feature fusion (ablation) | Accel-18 · Cityscapes, keyframe interval 5, max offset | inference time: 0.46 -> 0.44 | Helped |
| Use score fusion instead of feature fusion (ablation) | Accel-18 · Cityscapes, keyframe interval 5, max offset | mIoU: 69.5 -> 69.2 | Hurt |
| Use smaller intermediate channel dimensions (512 or 1024 rather than 2048) (ablation) | Accel · Segmentation experiments | segmentation accuracy | Hurt |
Closest papers by meaning
- Exploring Temporal Feature Correlation for Efficient and Stable Video Semantic Segmentation (AAAI 2024 · similarity 0.87)
- Exploiting Temporal State Space Sharing for Video Semantic Segmentation (CVPR 2025 · similarity 0.86)
- VidSeg: Training-free Video Semantic Segmentation based on Diffusion Models (CVPR 2025 · similarity 0.86)
- The Devil is in Temporal Token: High Quality Video Reasoning Segmentation (CVPR 2025 · similarity 0.86)
- Pay Attention to Target: Relation-Aware Temporal Consistency for Domain Adaptive Video Semantic Segmentation (AAAI 2024 · similarity 0.86)
- Semantic and Sequential Alignment for Referring Video Object Segmentation (CVPR 2025 · similarity 0.85)
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.