【交大云盘下载】【百度网盘下载 098k】【Google Drive download】

DAmageNet数据集的信息详见《Universal Adversarial Attack on Attention and the Resulting Dataset DAmageNet》

DAmageNet包含从ImageNet生成的大量可迁移对抗样本。

DAmageNet包含50000张224*224的图片，它们的原图被中心裁剪和调整大小。

DAmageNet中的图片，相比于原图约有7.32个像素值的均方根误差。

DAmageNet中的图片能欺骗在ImageNet上预训练好的模型，错误率可高达85%

DAmageNet中的图片能欺骗在ImageNet上对抗训练好的模型，错误率可高达70%

./DAmageNet中的文件和ILSVRC2012_img_val有着同样的名字。

可以使用附带的test.py测试DAmageNet

解压文件夹然后运行

python test.py DAmageNet VGG19,ResNet50,DenseNet121 0

使用本数据库请引用：

Sizhe Chen, Zhengbao He, Chengjin Sun, Jie Yang, Xiaolin Huang: Universal Adversarial Attack on Attention and the Resulting Dataset DamageNet, IEEE Transactions on Pattern Analysis and Machine Intelligence, doi:10.1109/TPAMI.2020.3033291

陈思哲，黄晓霖*，何正保，孙程锦

上海交通大学 图像处理与模式识别研究所

=======================================================================

Details of DAmageNet can be viewed in Universal Adversarial Attack on Attention and the Resulting Dataset DAmageNet

DAmageNet is a massive dataset containing universal adversarial samples generated from ImageNet.

DAmageNet contains 50000 224*224 images, whose original images have been centrally cropped and resized.

DAmageNet images have an average root mean square deviation of around 7.32 from original samples.

DAmageNet can fool pretrained models in ImageNet to have error rate up to 85%.

DAmageNet can fool adversarial-trained models in ImageNet to have error rate up to 70%.

Each file in DAmageNet has the same name as in ILSVRC2012_img_val.

Test in DAmageNet can be done by test.py

Unzip DAmageNet and run

python test.py DAmageNet VGG19,ResNet50,DenseNet121 0

For the reference: 

Sizhe Chen, Zhengbao He, Chengjin Sun, Jie Yang, Xiaolin Huang: Universal Adversarial Attack on Attention and the Resulting Dataset DamageNet, IEEE Transactions on Pattern Analysis and Machine Intelligence, doi:10.1109/TPAMI.2020.3033291

Sizhe Chen, Xiaolin Huang*, Zhengbao He, Chengjin Sun

Institute of Image Processing and Pattern Recognition,  Shanghai Jiao Tong University

【旧版交大云盘】【旧版百度网盘下载】【old version Google Drive download】