A simple and effective road semantic segmentation model, easy to use, very suitable for novices to get started.This repo is based on this Python blog.
Install OneFlow:
- oneflow 0.5.0.dev20210801+cu102
Install other requirements
$ pip install -r requirements.txt
Download required dataset from the following link: https://oneflow-public.oss-cn-beijing.aliyuncs.com/datasets/models/seg/FODDET/CamVid.zip
Download pretrained model from the following link: https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/cv/seg/FODDET/UNetmodel.zip
-
The training code
train.py
mainly contains data reading code, segmentation model code, and specific training code. -
The test code
test.py
mainly visualizes the segmentation effect, and the code content mainly includes the import of the trained model and the test process.You can download the trained model and use it directly.
$ bash train/train.sh
$ bash test/test.sh
The following shows some results of road image segmentation.