SH-Net

https://link.springer.com/article/10.1007/s10489-022-04014-0

Requirements

Install Pytorch.

Training data

The training data can be downloaed from (EORSSD) Dataset. Our edge map from: https://github.com/yun-liu/rcf.

Pretrained weights for backbone

The pretrianed weight for backbone can be downloaded from [VGG16].(BaiduNetDisk：https://pan.baidu.com/s/1hbIShLB-iYuRXy59iVvg8g extract code：msrp).

Test Model

1. Modify the testing image path accordingly.
2. You need to get the test images (BaiduNetDisk：https://pan.baidu.com/s/1u8W5iGMPFnQCdhaV1d6FHw extract code：6h6e).
3. Run ./test.py

Visualization

Code in visual.py helps generate visualization. You can find our results on EORSSD and ORSSD dataset here: BaiduNetDisk：https://pan.baidu.com/s/1DG35rlwtWV8RBJnNTnzALQ extract code：k9c4

You can find our visual results on EORSSD and ORSSD dataset here: BaiduNetDisk： https://pan.baidu.com/s/15d2r_EJ_YHCRrmSXpGrM2g extract code：n1ce

Acknowledgement

Many thanks to [Weakly-Supervised Salient Object Detection via Scribble Annotations] https://github.com/JingZhang617/Scribble_Saliency.