in

Offical implemention of Robust Superpixel-Guided Attentional Adversarial Attack (CVPR2020)


Offical implemention of Robust Superpixel-Guided Attentional Adversarial Attack (CVPR2020)

Setup

  • Install python — This repo is tested with 3.6

  • Install PyTorch version >= 1.0.0, torchvision >= 0.2.1

  • Download Inception-v3 pretrained model from https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth and move it to ./PRETRAIN/

Super-Pixel Generation

K: number of super pixels

M: compactness factor

We resize the image to 299×299 as pre-process and 400 images example can be found at here

Adversarial Attack

run on ImageNet dataset for white-box attack succes rate and robustness toward resize(2x)

GitHub

View Github


South Indian Bank Recruitment 2022 For Clerk, PO and Lateral PO Posts

Does Kaleido work on Google Colab (as Orca is crap)?