SEINet

This project provides the code and results for 'Semantic-Edge Interactive Network for Salient Object Detection in Optical Remote Sensing Images', IEEE JSTARS, 2023. IEEE link

Network Architecture

Requirements

python 3.7 + pytorch 1.9.0

Saliency maps

SEINet_EfficientNetB7_ORSI-4199 saliency maps: Baidu (code:SEIN)
SEINet_EfficientNetB7_EORSSD saliency maps: Baidu (code:SEIN)
SEINet_EfficientNetB7_ORSSD saliency maps: Baidu (code:SEIN)
SEINet_Res2Net50_ORSI-4199 saliency maps: Baidu (code:SEIN)
SEINet_Res2Net50_EORSSD saliency maps: Baidu (code:SEIN)
SEINet_Res2Net50_ORSSD saliency maps: Baidu (code:SEIN)
SEINet_ResNet50_ORSI-4199 saliency maps: Baidu (code:SEIN)
SEINet_ResNet50_EORSSD saliency maps: Baidu (code:SEIN)
SEINet_ResNet50_ORSSD saliency maps: Baidu (code:SEIN)
SEINet_VGG16_ORSI-4199 saliency maps: Baidu (code:SEIN)
SEINet_VGG16_EORSSD saliency maps: Baidu (code:SEIN)
SEINet_VGG16_ORSSD saliency maps: Baidu (code:SEIN)

Training

Run train_SEINet.py.
For SEINet_VGG16, please modify paths of VGG backbone (code: SEIN) in /model/vgg.py.

Pre-trained model and testing

Download the following pre-trained model and put them in ./models/, then run test_SEINet.py.
SEINet_EfficientNetB7_ORSI-4199 (code:SEIN)
SEINet_EfficientNetB7_EORSSD (code:SEIN)
SEINet_EfficientNetB7_ORSSD (code:SEIN)
SEINet_Res2Net50_ORSI-4199 (code:SEIN)
SEINet_Res2Net50_EORSSD (code:SEIN)
SEINet_Res2Net50_ORSSD (code:SEIN)
SEINet_ResNet50_ORSI-4199 (code:SEIN)
SEINet_ResNet50_EORSSD (code:SEIN)
SEINet_ResNet50_ORSSD (code:SEIN)
SEINet_VGG16_ORSI-4199 (code:SEIN)
SEINet_VGG16_EORSSD (code:SEIN)
SEINet_VGG16_ORSSD (code:SEIN)

Evaluation Tool

You can use the evaluation tool (MATLAB version) to evaluate the above saliency maps.

Acknowledgement

We would like to thank the contributors to the MCCNet.