WaLittleMoon / Learning-Rich-Features-for-Image-Manipulation-Detection

基于双流 Faster R-CNN 网络的 图像篡改检测
123 stars 26 forks source link

Learning-Rich-Features-for-Image-Manipulation-Detection

基于双流 Faster R-CNN 网络的 图像篡改检测

代码说明

本实验主要是对dBekerFaster-RCNN-TensorFlow-Python3GitHub仓库代码进行学习和一定的修改,从而实现了双流篡改检测。

参考链接:

https://github.com/dBeker/Faster-RCNN-TensorFlow-Python3

部署说明

首先修改_lib文件夹为lib

由于GitHub文件上传大小的限制,预处理网络和训练好的模型请从网盘下载: 链接: https://pan.baidu.com/s/1eav5wfVrHFzeP1xCsp_fsQ 提取码: pm8m

分享包括两个文件夹:

将vgg16网络文件夹下的.ckpt文件放在Learning-Rich-Features-for-Image-Manipulation-Detection/data/imagenet_weights/文件夹下; 将训练好的网络参数文件夹下的四个文件放在Learning-Rich-Features-for-Image-Manipulation-Detection/default/gene_2007_trainval/default/文件夹下。

然后直接运行双流Faster R-CNN.ipynb文件。

运行异常处理