Road Lane Segmentation
[](https://python3statement.org/#sections50-why)
[](https://github.com/psf/black)
[](https://pre-commit.com/)
[](https://python-poetry.org/)
This project focuses on the segmentation of city road lane markings using deep machine learning techniques.
.gif)
## Trained Models
- The latest:
best-valid-iou_1c07833020cf45939f24b5bec2bada5a.pth
- Trained in T204:
best-valid-iou_40446a2b3b7543c292301a3b2da1ed67.pth
T204 trained on datasets with errors
## Metrics
| Class label | Class name | IoU - The latest | IoU - T204 | The latest DICE | T204 DICE |
| :---------: | :------------------------: | :--------------: | :--------: | :-------------: | :-------: |
| background | background | 0.998 | 0.998 | 0.999 | 0.998 |
| SYD | solid yellow dividing line | 0.986 | 0.823 | 0.987 | 0.899 |
| BWG | broken white guiding line | 0.693 | 0.568 | 0.775 | 0.682 |
| SWD | solid white dividing line | 0.794 | 0.617 | 0.881 | 0.727 |
| SWS | solid white stop line | 0.114 | 0.432 | 0.162 | 0.54 |
| CWYZ | crosswalk | 0.386 | 0.56 | 0.496 | 0.687 |
| - | - | - | - | - | - |
| micro | | 0.995 | 0.995 | 0.998 | 0.998 |
| macro | | 0.662 | 0.667 | 0.717 | 0.756 |
## Articles
A. R. Muzalevskiy, E. V. Serykh, M. M. Kopichev, E. V. Druian and M. A. Chernyshev, "Lane Marking Semantic Segmentation Using Convolutional Neural Networks," 2023 XXVI International Conference on Soft Computing and Measurements (SCM), Saint Petersburg, Russian Federation, 2023, pp. 123-126, doi: 10.1109/SCM58628.2023.10159034.
## Repository contents
System Requirements
- Tested on the Linux Ubuntu 20.04 & 22.04
- Supports 1.2.0 <= poetry <= 1.8.2