Features:
- Reduced Anchor Requirements: 20 anchors is all you need.
- End-to-End Capability: Facilitates training and evaluation absent NMS post-processing.
- Deployment-friendly: Comprises solely of CNNs and MLPs.
- Scalable Framework: Encompasses data preprocessing, model training, performance assessment, and visualization.
 |
 |
 |
 |
 |
 |
 |
 |
For the preparation of datasets and environments, as well as detailed commands, please refer to INSTALL.md.
We provide trained model weights and corresponding config files for CULane, Tusimple, LLAMAS, DL-Rail, and CurveLanes.
| Dataset | Backbone | Performance (NMS-free) | Config | Weight-Link |
|---|---|---|---|---|
| CULane | ResNet18 | 80.81 (F1@50) | culane_r18 | polarrcnn_culane_r18.pth |
| CULane | ResNet34 | 80.92 (F1@50) | culane_r34 | polarrcnn_culane_r34.pth |
| CULane | ResNet50 | 81.34 (F1@50) | culane_r50 | polarrcnn_culane_r50.pth |
| CULane | DLA34 | 81.49 (F1@50) | culane_dla34 | polarrcnn_culane_dla34.pth |
| Tusimple | ResNet18 | 97.94 (F1) | tusimple_r18 | polarrcnn_tusimple_r18.pth |
| LLAMAS | ResNet18 | 96.06 (F1@50) | llamas_r18 | polarrcnn_llamas_r18.pth |
| LLAMAS | DLA34 | 96.14 (F1@50) | llamas_dla34 | polarrcnn_llamas_dla34.pth |
| DL-Rail | ResNet18 | 97.00 (F1@50) | dlrail_r18 | polarrcnn_dlrail_r18.pth |
| CurveLanes | DLA34 | 87.29 (F1@50) | curvelanes_dla34 | polarrcnn_curvelanes_dla34.pth |
@inproceedings{zheng2022clrnet,
title={Clrnet: Cross layer refinement network for lane detection},
author={Zheng, Tu and Huang, Yifei and Liu, Yang and Tang, Wenjian and Yang, Zheng and Cai, Deng and He, Xiaofei},
booktitle={Proceedings of the IEEE/CVF conference on computer vision and pattern recognition},
pages={898--907},
year={2022}
}
@inproceedings{honda2024clrernet,
title={CLRerNet: improving confidence of lane detection with LaneIoU},
author={Honda, Hiroto and Uchida, Yusuke},
booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
pages={1176--1185},
year={2024}
}
@inproceedings{chen2024sketch,
title={Sketch and Refine: Towards Fast and Accurate Lane Detection},
author={Chen, Chao and Liu, Jie and Zhou, Chang and Tang, Jie and Wu, Gangshan},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
volume={38},
number={2},
pages={1001--1009},
year={2024}
}