code for our paper A Deep Spatial Contextual Long-Term Recurrent Convolutional Network for Saliency Detection
Note: our network needs the pretrained ResNet50 and placesCNN models. For the latter, we changed its layer names to avoid the conflict with the ResNet layers. You can downnload it with our trained DSCLRCN model from Google drive or Baidu Yun.
Please cite our work if it helps your research:
@article{liu2018deep,
title={A Deep Spatial Contextual Long-Term Recurrent Convolutional Network for Saliency Detection},
author={Liu, Nian and Han, Junwei},
journal={IEEE Transactions on Image Processing},
volume={27},
number={7},
pages={3264--3274},
year={2018},
publisher={IEEE}
}