Deep learning models for classification of 15 common weeds in the southern U.S. cotton production systems.
- pytorch
- torchsummary
- tensorboard
- PIL
- Scikit-learn
- The weed image dataset is publicly available at https://www.kaggle.com/yuzhenlu/cottonweedid15
- To prepare your own dataset, you can run
python common/partition_imgs_Ubuntu.py
- To train the models, just specify the name of the models, and then run
python train.py
. - To test the images, just specify the name of the models, and then run
python test.py
. - To eval new data, just specify the name of the models, and then run
python eval.py
. - To visualize the training, run
tensorboard --logdir=runs
Detailed documentation of deep transfer learning for weed classification of the cotton weed dataset is given in our arXiv paper: https://arxiv.org/abs/2110.04960. If you use the dataset or models in a publication, please cite this paper.
@misc{chen2021performance,
title={Performance Evaluation of Deep Transfer Learning on Multiclass Identification of Common Weed Species in Cotton Production Systems},
author={Dong Chen, Yuzhen Lu, Zhaojiang Li, Sierra Young},
year={2021},
eprint={2110.04960},
archivePrefix={arXiv},
primaryClass={cs.CV}
}