wangyusenn / afrnet

AFRNet

This project provides the code and results for 'Transmission Line Detection Through Auxiliary Feature Registration with Knowledge Distillation'

Requirements

Python 3.7+, Pytorch 1.5.0+, Cuda 10.2+, TensorboardX 2.1, opencv-python If anything goes wrong with the environment, please check requirements.txt for details.

Architecture and Details

Results

Data Preparation

数据集： 数据集链接

Training & Testing

modify the train_root val_root save_path test_path in config.py according to your own data path.

Train the AFRNet:

python train.py

modify the test_path path in test.py according to your own data path.

Test the AFRNet:

python test.py

Evaluate tools

Implemented through evaluate.py, specific predictions can be obtained by setting the path inside.

Saliency Maps and Weights

百度网盘链接：Saliency Maps and Weights

Pretraining Models

百度网盘链接: Pretraining Models

Contact

Please drop me an email for any problems or discussion: https://wujiezhou.github.io/ ([email protected]) or [email protected]