Transferable Adversarial Attacks for Image and Video Object Detection

Published in

International Joint Conference on Artificial In...(2018)

External Links:

Generate Graph DownloadPDF

TL;DR

The proposed method is based on the Generative Adversarial Network (GAN) framework, where it combines the high-level class loss and low-level feature loss to jointly train the adversarial example generator, and can efficiently generate image and video adversarial examples that have better transferability.

Abstract

Identifying adversarial examples is beneficial for understanding deep networks and developing robust models. However, existing attacking methods for image object detection have two limitations: weak transferability---the generated adversarial examples often have a low success rate to attack other kinds of detection methods, and high computation cost---they need much time to deal with video data, where many frames need polluting. To address these issues, we present a generative method to obtain adversarial images and videos, thereby significantly reducing the processing time. To enhance transferability, we manipulate the feature maps extracted by a feature network, which usually constitutes the basis of object detectors. Our method is based on the Generative Adversarial Network (GAN) framework, where we combine a high-level class loss and a low-level feature loss to jointly train the adversarial example generator. Experimental results on PASCAL VOC and ImageNet VID datasets show that our method efficiently generates image and video adversarial examples, and more importantly, these adversarial examples have better transferability, therefore being able to simultaneously attack two kinds of representative object detection models: proposal based models like Faster-RCNN and regression based models like SSD.

Authors

Xiaochun Cao

3 papers

Jun Zhu

7 papers

Xingxing Wei

3 papers

References45 items

Volume 1

Video-to-Video Translation with Global Temporal Consistency

Robust Adversarial Perturbation on Deep Proposal-based Models

Textbook Question Answering Under Instructor Guidance with Memory Networks

Adversarial Attacks on Face Detectors Using Neural Net Based Constrained Optimization

Transferable Adversarial Attacks for Image and Video Object Detection

Published in

International Joint Conference on Artificial In...(2018)

External Links:

Generate Graph DownloadPDF

TL;DR

Abstract

Authors

Xiaochun Cao

3 papers

Jun Zhu

7 papers

Xingxing Wei

3 papers

References45 items

Volume 1

Video-to-Video Translation with Global Temporal Consistency

Robust Adversarial Perturbation on Deep Proposal-based Models

Textbook Question Answering Under Instructor Guidance with Memory Networks

Adversarial Attacks on Face Detectors Using Neural Net Based Constrained Optimization

Siyuan Liang

1 papers

ShapeShifter: Robust Physical Adversarial Attack on Faster R-CNN Object Detector

Optimizing Video Object Detection via a Scale-Time Lattice

Robust Physical Adversarial Attack on Faster R-CNN Object Detector

Sparse Adversarial Perturbations for Videos

Certified Defenses against Adversarial Examples

Generating Adversarial Examples with Adversarial Networks

Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey

Towards High Performance Video Object Detection

Boosting Adversarial Attacks with Momentum

Towards Interpretable Deep Neural Networks by Leveraging Adversarial Examples

Robust Physical-World Attacks on Deep Learning Models

Towards Robust Detection of Adversarial Examples

Flow-Guided Feature Aggregation for Video Object Detection

Adversarial Examples for Semantic Segmentation and Object Detection

Deep Feature Flow for Video Recognition

Image-to-Image Translation with Conditional Adversarial Networks

Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition

Towards Evaluating the Robustness of Neural Networks

R-FCN: Object Detection via Region-based Fully Convolutional Networks

SSD: Single Shot MultiBox Detector

DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks

You Only Look Once: Unified, Real-Time Object Detection

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Adam: A Method for Stochastic Optimization

Explaining and Harnessing Adversarial Examples

Deep visual-semantic alignments for generating image descriptions

Very Deep Convolutional Networks for Large-Scale Image Recognition

Intriguing properties of neural networks

Joint Deep Learning for Pedestrian Detection

Visualizing and Understanding Convolutional Networks

IEEE Transactions on Pattern Analysis and Machine Intelligence

In 2017 IEEE Symposium on Security and Privacy

Region-Based Convolutional Networks for Accurate Object Detection and Segmentation

The PASCAL Visual Object Classes Challenge

The Pascal Visual Object Classes Challenge 2006 ( VOC 2006 ) Results

Field of Study

Computer Science

Journal Information

Name

ArXiv

Volume

abs/2005.00687

Venue Information

Name

International Joint Conference on Artificial Intelligence

Type

conference

URL

http://www.ijcai.org/

Alternate Names

Int Jt Conf Artif Intell
IJCAI

TL;DR

Abstract

Authors

References45 items

Volume 1

Video-to-Video Translation with Global Temporal Consistency

Robust Adversarial Perturbation on Deep Proposal-based Models

Textbook Question Answering Under Instructor Guidance with Memory Networks

Adversarial Attacks on Face Detectors Using Neural Net Based Constrained Optimization

TL;DR

Abstract

Authors

References45 items

Volume 1

Video-to-Video Translation with Global Temporal Consistency

Robust Adversarial Perturbation on Deep Proposal-based Models

Textbook Question Answering Under Instructor Guidance with Memory Networks

Adversarial Attacks on Face Detectors Using Neural Net Based Constrained Optimization

ShapeShifter: Robust Physical Adversarial Attack on Faster R-CNN Object Detector

Optimizing Video Object Detection via a Scale-Time Lattice

Robust Physical Adversarial Attack on Faster R-CNN Object Detector

Sparse Adversarial Perturbations for Videos

Certified Defenses against Adversarial Examples

Generating Adversarial Examples with Adversarial Networks

Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey

Towards High Performance Video Object Detection

Boosting Adversarial Attacks with Momentum

Towards Interpretable Deep Neural Networks by Leveraging Adversarial Examples

Robust Physical-World Attacks on Deep Learning Models

Towards Robust Detection of Adversarial Examples

Flow-Guided Feature Aggregation for Video Object Detection

Adversarial Examples for Semantic Segmentation and Object Detection

Mask R-CNN

Deep Feature Flow for Video Recognition

Image-to-Image Translation with Conditional Adversarial Networks

Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition

Towards Evaluating the Robustness of Neural Networks

R-FCN: Object Detection via Region-based Fully Convolutional Networks

SSD: Single Shot MultiBox Detector

DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks

You Only Look Once: Unified, Real-Time Object Detection

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Adam: A Method for Stochastic Optimization

Explaining and Harnessing Adversarial Examples

Deep visual-semantic alignments for generating image descriptions

Very Deep Convolutional Networks for Large-Scale Image Recognition

Intriguing properties of neural networks

Joint Deep Learning for Pedestrian Detection

Visualizing and Understanding Convolutional Networks

Et al

IEEE Transactions on Pattern Analysis and Machine Intelligence

In NeurIPS

In CVPR

In ICCV

In 2017 IEEE Symposium on Security and Privacy

Region-Based Convolutional Networks for Accurate Object Detection and Segmentation

The PASCAL Visual Object Classes Challenge

The Pascal Visual Object Classes Challenge 2006 ( VOC 2006 ) Results

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names