In recent years, there has been increasing interest in studying the adversarial attack, which poses potential risks to deep learning applications and has stimulated numerous researches, e.g. improving the robustness of deep neural networks. In this work, we propose a novel double-stream architecture – Guided Erasable Adversarial Attack (GEAA) – for protecting high-quality labeled data with high commercial values under data-sharing scenarios. GEAA contains three phases, the double-stream adversarial attack, denoising reconstruction, and watermark extraction. Specifically, the double-stream adversarial attack injects erasable perturbations into the training data to avoid database abuse. The denoising reconstruction rebuilds the traceable denoising data from adversarial examples. The watermark extraction recovers identity information from the denoised data for copyright protection. Additionally, we introduce the annealing optimization strategy to balance these phases and a boundary constraint to degrade the availability of adversarial examples. Through extensive experiments, we demonstrate the effectiveness of the proposed framework in data protection. The Pytorch ® implementations of GEAA can be downloaded from an open-source Github project https://github.com/Dlut-lab-zmn/GEAA-for-data-protection .
Recently, fully convolutional neural networks have been adopted for salient object detection and object contour detection, and have achieved impressive performance. Closely related contours are employed to help supervise low-level features, rather than being simultaneously trained as associated tasks as in most methods. This study proposes a coarse-to-fine architecture for a contour-aware recurrent cross constraint network (CARCCNet) for salient object detection. At the coarse stage, we design a contour-aware recurrent constraint network (CARCNet) with a recurrent structure that consists of a set of contour-aware constraint modules (CACMs), saliency-aware constraint modules (SACMs), and double supervised prediction modules (DSPMs). These modules can simultaneously generate saliency maps and contour maps and alternately constrain them at each recurrent step. In the refining stage, we propose a contour knowledge transfer residual (CKTR) module to transfer the contour knowledge from the low-level branch into the saliency features to obtain the final saliency map with complete objects and accurate contours. Our CARCCNet also finally generates the object contour map at the same time without post-processing. Extensive experiments on five saliency detection benchmark datasets demonstrate the effectiveness and robustness of the proposed method.
Scene graph generation (SGG) effectively improves semantic understanding of the visual world. However, the recent interest of researchers focuses on enhancing SGG in non-adversarial settings, which raises our curiosity about the adversarial robustness of SGG models. To bridge this gap, we perform adversarial attacks on two typical SGG tasks, Scene Graph Detection (SGDet) and Scene Graph Classification (SGCls). Specifically, we initially propose a bounding box relabeling method to reconstruct reasonable attack targets for SGCls. It solves the inconsistency between the specified bounding boxes and the scene graphs selected as attack targets. Subsequently, we introduce a two-step weighted attack by removing the predicted objects and relational triples that affect attack performance, which significantly increases the success rate of adversarial attacks on two SGG tasks. Extensive experiments demonstrate the effectiveness of our methods on five popular SGG models and four adversarial attacks. The Pytorch® implementation can be downloaded from an open-source Github project https://github.com/Dlut-lab-zmn/SGG_Attack.