Towards Domain Generalization in Crop and Weed Segmentation for Precision Farming Robots
We present a novel approach to leverage unlabeled images captured from various agricultural fields to develop domain generalized CNNs that enables agricultural robots to perform a reliable semantic segmentation of the classes soil, crop, and weed in different fields.
conda create -n dgcws python=3.8
conda activate dgcws
pip install -r ./requirements.txt
pip install torch==1.10.1+cu113 torchvision==0.11.2+cu113 torchaudio==0.10.1+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
pip install setuptools==59.5.0
Please note that the cuda version depends on your local machine.
python train.py --config ./configs/erfnet/config_uav_bonn.yaml --export_dir </path/to/export/directory>
Before you start the training you need to specify the paths to the datasets in the corresponding configuration file.
python test.py --config ./configs/erfnet/config_uav_bonn.yaml --export_dir </path/to/export/directory> --ckpt_path <path/to/erfnet.ckpt>
Before you start the testing you need to specify the paths to the datasets in the corresponding configuration file.
This software is released under a creative commons license which allows for personal and research use only. For a commercial license please contact the authors. You can view a license summary here.