TCLNet: Learning to Locate Typhoon Center using Deep Neural Network

This repository contains the source code, dataset and pretrained model for TCLNet, provided by Chao Tan.

The paper is avaliable for download here. Click here for more details.

---

Dataset & Pretrained Model

TCLD (Typhoon Center Location Dataset) is a brand new typhoon center location dataset for deep learning research. It contains 1809 grayscale images for training and another 319 images for testing. Our TCLD dataset is available for download at TianYiCloud(600MB) or BaiduCloud(600MB)(extraction code: ixmb).
You can get the TCLD dataset at any time but only for scientific research. At the same time, please cite our work when you use the TCLD dataset.

The pretrained model of our TCLNet on TCLD dataset can be download at TianYiCloud or BaiduCloud (extraction code: 8l1j).

Prerequisites

• Python 3.7
• PyTorch >= 1.4.0
• opencv 0.4
• PyQt 4
• numpy
• visdom

Training

1. Please download and unzip TCLD dataset and place it in datasets/data folder.
2. Run python -m visdom.server" to activate visdom server.
3. Run python run.py to start training from scratch.
4. You can easily monitor training process at any time by visiting http://localhost:8097 in your browser.

Testing

1. For TCLD dataset, please download and unzip pretrained model and place it in checkpoints folder.
2. Replace the test data in the daatsets/data/TCLD/TEST_INPUT" folder with your own data (optional).
3. Run python test.py to start testing.
4. The results of the testing will be saved in the checkpoint/TCLNET/evaluation" directory.

Citation

@inproceedings{
     title={{TCLNet}: Learning to Locate Typhoon Center Using Deep Neural Network},
     author={Tan, Chao},
     booktitle={IGARSS},
     year={2021},
     note={to appear},
}