Code for "Adversarial attack by dropping information." (ICCV 2021)

Last update: Nov 10, 2022

Overview

AdvDrop

Code for "AdvDrop: Adversarial Attack to DNNs by Dropping Information(ICCV 2021)."

Human can easily recognize visual objects with lost information: even losing most details with only contour reserved, e.g. cartoon. However, in terms of visual perception of Deep Neural Networks (DNNs), the ability for recognizing abstract objects (visual item with lost information) is still a challenge. In this work, we investigate this issue from an adversarial viewpoint: will the performance of DNNs decrease even for the images only losing a little information? Towards this end, we propose a novel adversarial attack, named AdvDrop, which crafts adversarial examples by dropping existing information of images. Previously, most adversarial attacks add extra disturbing information on clean images explicitly. Opposite to previous works, our proposed work explores the adversarial robustness of DNN models in a novel perspective by dropping imperceptible details to craft adversarial examples.

[Paper link](http://arxiv.org/abs/2108.09034)

Installation

We highly recommend using conda.

conda create -n advdrop_env python=3.7
source activate advdrop_env

After activating the virtual environment, install pytorch, numpy and torchattacks with:

pip install --user [the package name]

Download dataset.

Usage

Quick Start

python infod_sample.py

Parameters can be speicified in infod_sample.py.

An example

Adv Images

Acknowledgments

We refered Michael's code for JPEG (torch version) part.
We referred [Harry's code style] (https://github.com/Harry24k/adversarial-attacks-pytorch)

Code for "Adversarial attack by dropping information." (ICCV 2021)

Related tags

Overview

AdvDrop

Installation

Usage

Quick Start

An example

Adv Images

Acknowledgments

Citation

Owner

Ranjie Duan

CondLaneNet: a Top-to-down Lane Detection Framework Based on Conditional Convolution

PiCIE: Unsupervised Semantic Segmentation using Invariance and Equivariance in clustering (CVPR2021)

Stereo Radiance Fields (SRF): Learning View Synthesis for Sparse Views of Novel Scenes

2021 CCF BDCI 全国信息检索挑战杯（CCIR-Cup）智能人机交互自然语言理解赛道第二名参赛解决方案

Building blocks for uncertainty-aware cycle consistency presented at NeurIPS'21.

Artificial Neural network regression model to predict the energy output in a combined cycle power plant.

FAIR's research platform for object detection research, implementing popular algorithms like Mask R-CNN and RetinaNet.

Code for "Hierarchical Skills for Efficient Exploration" HSD-3 Algorithm and Baselines

这是一个mobilenet-yolov4-lite的库，把yolov4主干网络修改成了mobilenet，修改了Panet的卷积组成，使参数量大幅度缩小。

Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation

Collections for the lasted paper about multi-view clustering methods (papers, codes)

iris - Open Source Photos Platform Powered by PyTorch

CS50x-AI - Artificial Intelligence with Python from Harvard University

for a paper about leveraging discourse markers for training new models

A library for optimization on Riemannian manifolds

Semi-Supervised Learning, Object Detection, ICCV2021

Image-retrieval-baseline - MUGE Multimodal Retrieval Baseline

This is a simple framework to make object detection dataset very quickly

UnFlow: Unsupervised Learning of Optical Flow with a Bidirectional Census Loss

Adversarial-Information-Bottleneck - Distilling Robust and Non-Robust Features in Adversarial Examples by Information Bottleneck (NeurIPS21)