This is the source code of paper FusionLane: Multi-Sensor Fusion for Lane Marking Semantic Segmentation Using Deep Neural Networks. The source code of this project is modified based on deeplabV3+.
Requirements:
- Tensorflow>= 1.12
- Numpy
- matplotlib
- opencv-python
This project uses the TFRecord format to consume data in the training and evaluation process. The TFRecord used in this project could is available at here. You can also find the testing images at here.
You can start your own training from scratch as follows. Or you could just modified in the train.py file.
python train.py -- data_dir TFRECORD_DATA_DIR \
-- model_dir MODEL_DIR
You can evaluate you model in the same way as the training. Our model is available at here.
The rests are still under construction... O(∩_∩)O