Research map / Detection and segmentation / CNN features
Rethinking BiSeNet for Real-Time Semantic Segmentation
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 |
|---|---|---|---|
| Use STDC2 instead of GhostNet as the backbone (ablation) | STDC2-Seg · Cityscapes val, 512 × 1024 | mIoU: 67.8 -> 74.2 | Helped |
| Use STDC2 instead of MobileNetV3 as the backbone (ablation) | STDC2-Seg · Cityscapes val, 512 × 1024 | mIoU: 70.1 -> 74.2 | Helped |
| Use STDC2 instead of EfficientNet-B0 as the backbone (ablation) | STDC2-Seg · Cityscapes val, 512 × 1024 | mIoU: 72.2 -> 74.2 | Helped |
| Use STDC2 instead of EfficientNet-B0 as the backbone (ablation) | STDC2-Seg · Cityscapes val, 768 × 1536 | FPS: 45.9 -> 97.0 | Helped |
| Add Detail Guidance at 1x, 2x, and 4x (ablation) | STDC2-Seg50 · Cityscapes val | mIoU: 73.0 -> 74.2 | Helped |
| Add Detail Guidance at 1x, 2x, and 4x (ablation) | STDC2-Seg50 · Cityscapes val | FPS: 188.6 -> 188.6 | No clear effect |
| Add a Spatial Path instead of using Detail Guidance (ablation) | STDC2-Seg50 · Cityscapes val | mIoU: 73.0 -> 73.7 | Helped |
| Add a Spatial Path instead of using Detail Guidance (ablation) | STDC2-Seg50 · Cityscapes val | FPS: 188.6 -> 171.6 | Hurt |
| Use Detail Guidance at 1x, 2x, and 4x rather than a single detail scale (ablation) | STDC2-Seg50 · Cityscapes val | mIoU: 73.8 -> 74.2 | Helped |
| Increase the STDC module block count to 4 (ablation) | STDC2 · ImageNet | accuracy | Helped |
Closest papers by meaning
- DFANet: Deep Feature Aggregation for Real-Time Semantic Segmentation (CVPR 2019 · similarity 0.88)
- PIDNet: A Real-Time Semantic Segmentation Network Inspired by PID Controllers (CVPR 2023 · similarity 0.87)
- SCTNet: Single-Branch CNN with Transformer Semantic Information for Real-Time Segmentation (AAAI 2024 · similarity 0.86)
- SasWOT: Real-Time Semantic Segmentation Architecture Search WithOut Training (AAAI 2024 · similarity 0.85)
- Multi-view Aggregation Network for Dichotomous Image Segmentation (CVPR 2024 · similarity 0.85)
- Learning Hierarchical Image Segmentation For Recognition and By Recognition (ICLR 2024 · 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.