• 设为首页
  • 点击收藏
  • 手机版
    手机扫一扫访问
    迪恩网络手机版
  • 关注官方公众号
    微信扫一扫关注
    公众号

jawadbappy/forgery_localization_HLED: Implementation of "Hybrid LSTM and En ...

原作者: [db:作者] 来自: 网络 收藏 邀请

开源软件名称(OpenSource Name):

jawadbappy/forgery_localization_HLED

开源软件地址(OpenSource Url):

https://github.com/jawadbappy/forgery_localization_HLED

开源编程语言(OpenSource Language):

C 64.8%

开源软件介绍(OpenSource Introduction):

Localization of Image Forgeries

This project presents a framework to localize the image manipulation from an image. The network employs resampling features, Long-Short Term Memory (LSTM) cells, and encoder-decoder network in order to segment out manipulated regions from an image.

Data

We create a large dataset by splicing different objects obtained from MS-COCO dataset into the images of DRESDEN benchmark. Please check "synthetic_data" folder for more details.

Model

Model can be found in "./model" folder. Please note that the given model is obtained by finetuning the base model with NIST data. The base model is trained on synthesized data.

Resampling Features

The codes for extracting resampling features can be found on "Radon" folder. Please change the input and output directory for your own dataset. Following is the command to extract the resampling features.

python extract_resamp_feat.py

In this code, the images are stored in hdf5 format. Please note that the package "pyfftw" are required to be installed before running the script. Please use the following command to install the package.

sudo pip install pyfftw

Train

First, the data needs to be prepared either hdf5 format or any other formats. The training code needs to be modified accordingly. In order to train the model, an image and a corresponding binary mask is required.

python train.py

The model will be stored in the model path.

Test

We provide 8 sample images which will be found on test_data folder in order to demonstrate the output of the network. Please use the following command.

python test.py

The code will automatically generate the binary mask and the heat map of prediction score.

Sample Outputs

Screenshot Screenshot

Citation

Please cite the following paper for reference.

@article{bappy2019hybrid,
  title={Hybrid LSTM and Encoder-Decoder Architecture for Detection of Image Forgeries},
  author={Bappy, Jawadul H and Simons, Cody and Nataraj, Lakshmanan and Manjunath, BS and Roy-Chowdhury, Amit K},
  journal={IEEE Transactions on Image Processing},
  year={2019},
  publisher={IEEE}
}



鲜花

握手

雷人

路过

鸡蛋
该文章已有0人参与评论

请发表评论

全部评论

专题导读
热门推荐
阅读排行榜

扫描微信二维码

查看手机版网站

随时了解更新最新资讯

139-2527-9053

在线客服(服务时间 9:00~18:00)

在线QQ客服
地址:深圳市南山区西丽大学城创智工业园
电邮:jeky_zhao#qq.com
移动电话:139-2527-9053

Powered by 互联科技 X3.4© 2001-2213 极客世界.|Sitemap