This repository contains virus dataset and code (Virus-Instance-Segmentation-Transformer-Network) , provided by Hainan University, Chinese Center for Disease Control and Prevention and Huazhong Agricultural University.

Dataset

The virus dataset consists of 622 images with a size of 900 * 820 of three similar viruses (SARS-CoV-2, FLUAV, RSV, HSV1, HAdV5 and VACV), with a total number of 3195 virus particles, which were labelled by virologists.

Environment

The code is developed and tested under the following configurations.

CUDA>=11.0, Python>=3.7, torch>=1.11.0

The weights of swin transformer can be download in： https://pan.baidu.com/s/105tIfA1IXsUtcBWCkeZYrQ?pwd=5smq, code：5smq

How to use

Adjust the weight and data path in "config.py" and run the code "train_Yolact_swin.py" and "eval.py".