Paper: CVPR2018, Learning Rich Features for Image Manipulation Detection
Code based on Faster-RCNN
Python 3.6
pytorch 1.3.1
resnet101_caffe.pth
Casia 1.0
The name of the files in Sp may be something wrong. Use code in faster-rcnn/data/VOC2013 to modify the name and check it.
Thanks to @namtpham to offer groudtruth for CASIA 1.0 and CASIA 2.0.
The principle to rename files in origin dataset is to match them to GT.
Because the input of faster-rcnn is in the formation of VOC2007 and VOV2012, so I change the Dataset to this formation.