Research map / Detection and segmentation / CNN features
Importance-Aware Semantic Segmentation in Self-Driving with Discrete Wasserstein Training
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 |
|---|---|---|---|
| Replace CE loss with Wasserstein linear ground metric | SegNet · Cityscapes Group 4 | mIoU: 57.0 -> 67.4 | Helped |
| Replace CE loss with IAL | SegNet · Cityscapes Group 4 | mIoU: 57.0 -> 65.7 | Helped |
| Use squared rather than linear ground metric (ablation) | Wasserstein SegNet · Cityscapes Group 4 | mIoU: 67.4 -> 68.0 | Helped |
| Use Huber rather than linear ground metric (ablation) | Wasserstein SegNet · Cityscapes Group 4 | mIoU: 67.4 -> 67.9 | Helped |
| Replace CE loss with Wasserstein linear ground metric | ENet · Cityscapes Group 4 | mIoU: 58.3 -> 69.1 | Helped |
| Use squared rather than linear ground metric (ablation) | Wasserstein ENet · Cityscapes Group 4 | mIoU: 69.1 -> 69.3 | Helped |
| Use Huber rather than linear ground metric (ablation) | Wasserstein ENet · Cityscapes Group 4 | mIoU: 69.1 -> 69.5 | Helped |
| Replace CE loss with Wasserstein linear ground metric | FCN · CamVid Groups 3/4 | mIoU: 69.6 -> 72.4 | Helped |
| Use squared rather than linear ground metric (ablation) | Wasserstein FCN · CamVid Groups 3/4 | mIoU: 72.4 -> 72.8 | Helped |
| Use Huber rather than linear ground metric (ablation) | Wasserstein FCN · CamVid Groups 3/4 | mIoU: 72.4 -> 72.8 | Helped |
| Replace LRENT with Wasserstein linear ground metric | DeeplabV2 · GTA5→Cityscapes Group 4 | mIoU: 46.5 -> 46.8 | Helped |
| Use squared rather than linear ground metric (ablation) | Wasserstein DeeplabV2 · GTA5→Cityscapes Group 4 | mIoU: 46.8 -> 47.0 | Helped |
Closest papers by meaning
- Severity-Aware Semantic Segmentation With Reinforced Wasserstein Training (CVPR 2020 · similarity 0.94)
- Cross-Dataset Collaborative Learning for Semantic Segmentation in Autonomous Driving (AAAI 2022 · similarity 0.89)
- Vanishing-Point-Guided Video Semantic Segmentation of Driving Scenes (CVPR 2024 · similarity 0.85)
- Weakly Supervised Semantic Segmentation for Driving Scenes (AAAI 2024 · similarity 0.84)
- Enhancing End-to-End Autonomous Driving with Latent World Model (ICLR 2025 · similarity 0.84)
- Generative Map Priors for Collaborative BEV Semantic Segmentation (CVPR 2025 · similarity 0.83)
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.