Can You Spot the Chameleon? Adversarially Camouflaging Images from Co-Salient Object Detection (2020-09-19T00:00:00.000000Z)

TL;DR

The very first blackbox joint adversarial exposure and noise attack (Jadena), where they jointly and locally tune the exposure and additive perturbations of the image according to a newly designed high-feature-level contrast-sensitive loss function, leads to significant performance degradation on various co- saliency detection datasets and makes the co-salient objects undetectable.

Abstract

Co-salient object detection (CoSOD) has recently achieved significant progress and played a key role in retrieval-related tasks. However, it inevitably poses an entirely new safety and security issue, i.e., highly personal and sensitive content can potentially be extracting by powerful CoSOD methods. In this paper, we address this problem from the perspective of adversarial attacks and identify a novel task: adversarial co-saliency attack. Specially, given an image selected from a group of images containing some common and salient objects, we aim to generate an adversarial version that can mislead CoSOD methods to predict incorrect co-salient regions. Note that, compared with general white-box adversarial attacks for classification, this new task faces two additional challenges: (1) low success rate due to the diverse appearance of images in the group; (2) low transferability across CoSOD methods due to the considerable difference between CoSOD pipelines. To address these challenges, we propose the very first blackbox joint adversarial exposure and noise attack (Jadena), where we jointly and locally tune the exposure and additive perturbations of the image according to a newly designed high-feature-level contrast-sensitive loss function. Our method, without any information on the state-of-the-art CoSOD methods, leads to significant performance degradation on various co-saliency detection datasets and makes the co-salient objects undetectable. This can have strong practical benefits in properly securing the large number of personal photos currently shared on the Internet. Moreover, our method is potential to be utilized as a metric for evaluating the robustness of CoSOD methods.

References63 items

Adversarial Relighting Against Face Recognition

Learning to Adversarially Blur Visual Object Tracking

AVA: Adversarial Vignetting Attack against Visual Recognition

AdvHaze: Adversarial Haze Attack

Fooling LiDAR Perception via Adversarial Trajectory Perturbation

Bias Field Poses a Threat to DNN-Based X-Ray Recognition

It's Raining Cats or Dogs? Adversarial Rain Attack on DNN Perception

Adversarial Exposure Attack on Diabetic Retinopathy Imagery

Adv-watermark: A Novel Watermark Perturbation for Adversarial Examples

Pasadena: Perceptually Aware and Stealthy Adversarial Denoise Attack

Re-Thinking Co-Salient Object Detection

Taking a Deeper Look at Co-Salient Object Detection

U2-Net: Going Deeper with Nested U-Structure for Salient Object Detection

Gradient-Induced Co-Saliency Detection

Adaptive Graph Convolutional Network With Attention Graph Clustering for Co-Saliency Detection

SAD: Saliency-based Defenses Against Adversarial Examples

Robust Deep Co-Saliency Detection With Group Semantic and Pyramid Attention

Watch out! Motion is Blurring the Vision of Your Deep Neural Networks

ABBA: Saliency-Regularized Motion-Based Adversarial Blur Attack

Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement

Amora: Black-box Adversarial Morphing Attack

ColorFool: Semantic Adversarial Colorization

SPARK: Spatial-Aware Online Incremental Attack Against Visual Tracking

A Unified Multiple Graph Learning and Convolutional Network Model for Co-saliency Estimation

EGNet: Edge Guidance Network for Salient Object Detection

Deep High-Resolution Representation Learning for Visual Recognition

Detecting Robust Co-Saliency with Recurrent Co-Attention Neural Network

Enhancing Underexposed Photos Using Perceptually Bidirectional Similarity

BASNet: Boundary-Aware Salient Object Detection

A Simple Pooling-Based Design for Real-Time Salient Object Detection

Salient Object Detection in the Deep Learning Era: An In-Depth Survey

Evading Defenses to Transferable Adversarial Examples by Translation-Invariant Attacks

Adversarial Attacks against Deep Saliency Models

Review of Visual Saliency Detection With Comprehensive Information

A Review of Co-Saliency Detection Algorithms

Quality-Guided Fusion-Based Co-Saliency Estimation for Image Co-Segmentation and Colocalization

Defense Against Adversarial Attacks Using High-Level Representation Guided Denoiser

Boosting Adversarial Attacks with Momentum

Learning to Detect Salient Objects with Image-Level Supervision

Saliency Detection with Recurrent Fully Convolutional Networks

Towards Evaluating the Robustness of Neural Networks

Adversarial examples in the physical world

A Weighted Variational Model for Simultaneous Reflectance and Illumination Estimation

Co-saliency detection via inter and intra saliency propagation

Detection of Co-salient Objects by Looking Deep and Wide

Deep Contrast Learning for Salient Object Detection

Deep Residual Learning for Image Recognition

A Self-Paced Multiple-Instance Learning Framework for Co-Saliency Detection

Multi-Camera Saliency

Complex Event Detection using Semantic Saliency and Nearly-Isotonic SVM

Visual saliency based on multiscale deep features

Beyond pixels: A comprehensive survey from bottom-up to semantic image segmentation and cosegmentation

Explaining and Harnessing Adversarial Examples

Co-localization in Real-World Images

Self-Adaptively Weighted Co-Saliency Detection via Rank Constraint

Repairing Bad Co-Segmentation Using Its Quality Evaluation and Segment Propagation