Please copy lib
dir in faster_rcnn.pytorch to your root
dir and install faster_rcnn.pytorch,
following the Preparation steps. Prepare the dataset of Pascal VOC according to faster_rcnn.pytorch. Please prepare SALICON dataset according to EML-Net.
Please refer to train2.py
and test_net.py
. Change the predefined parameters according to your own data path, model path, etc. Here we do the joint
learning of object detection and saliency prediction. While in the test stage, we only output saliency result. We use the Resnet50 backbone in the code.
The code is based on faster_rcnn.pytorch and EML-Net. Hope the code is helpful for the joint learning in the field of saliency prediction.