Search

• Input gradients of adversarially-trained robust models show a preferred direction. • These gradients point more directly to the closest point of inaccurate classes. • Generative adversarial networks estimate this direction and its alignment metric. • Metric correlates to robustness for models trained with projected gradient descent. • Enforcing the proposed alignment increases robustness of models. Adversarial training, especially projected gradient descent (PGD), has proven to be a successful approach for improving robustness against adversarial attacks. After adversarial training, gradients of models with respect to their inputs have a preferential direction. However, the direction of alignment is not mathematically well established, making it difficult to evaluate quantitatively. We propose a novel definition of this direction as the direction of the vector pointing toward the closest point of the support of the closest inaccurate class in decision space. To evaluate the alignment with this direction after adversarial training, we apply a metric that uses generative adversarial networks to produce the smallest residual needed to change the class present in the image. We show that PGD-trained models have a higher alignment than the baseline according to our definition, that our metric presents higher alignment values than a competing metric formulation, and that enforcing this alignment increases the robustness of models. [ABSTRACT FROM AUTHOR]

In this paper, we propose a Newton-based algorithm for nonnegative matrix factorization in image processing. We employ the new algorithm to three real-world databases. Extensive numerical results show the feasibility and validity of the proposed algorithm. [ABSTRACT FROM AUTHOR]

• A new gradient-direction-based method is proposed to avoid the denoised edges to be blurred especially when the edges of the guidance image are weak or inexistent. • The reconstructed gradient vectors are used for the purpose of making the guidance image deeply participate in the model optimization process. • A specifically designed optimization procedure is proposed to solve these nonconvex subproblems. • A new regularization term is formulated to weaken the effects of the unreliable prior information from the guidance image. • Experimental results on public datasets and from benchmark methods consistently demonstrate the effectiveness of the proposed method both visually and quantitatively. Joint image denoising algorithms use the structures of the guidance image as a prior to restore the noisy target image. While the provided guidance images are helpful to improve the denoising performance, the denoised edges are most likely to be blurred especially when the edges of the guidance image are weak or inexistent. To address this weakness, this paper proposes a new gradient-direction-based joint image denoising method in which the absolute cosine value of the angle between two gradient vectors of the guidance image and those of the image to recover is employed as the parallel measurement to ensure that the gradient directions of the denoised image are approximately the same as or opposite to those of the guidance image. Besides, a new edge-preserving regularization term is developed to alleviate the effects of the unreliable prior information from guidance image. To simplify the resultant complex nonconvex and nonlinear fractional model, the logarithm function is employed to convert the multiplication operation into addition operation. Then, we construct the surrogate function for the logarithmic term of l 2 -norm, and separate the variables to transform the objective function into convex one with high numerical stability while retaining high efficiency. Finally, the optimal solutions can be obtained by directly minimizing the convex functions. Experimental results on public datasets and from nine benchmark methods consistently demonstrate the effectiveness of the proposed method both visually and quantitatively. [ABSTRACT FROM AUTHOR]

Abstract: At present there are many methods that could deal well with frontal view face recognition. However, most of them cannot work well when there is only single example image per person. In order to deal with this problem of single example image per person is stored in the system in the real-world application. In this paper, a comparative study of the AFMT, Fourier-AFMT and Taylor-AFMT are identified for face recognition. And then, we present hybrid Fourier-AFMT framework based face recognition for feature extraction. Firstly, both directionality of edges and intensity facial features are extracted and secondly fuse two kinds of features and classify with correlation coefficient method (CCM). Experiments are implemented on YALE and ORL face databases to demonstrate the efficient of proposed methods. The experimental results show that the average recognition accuracy rates of our proposed fuse multiple feature domains much higher than that of single feature domain. [ABSTRACT FROM AUTHOR]

Face detection from cluttered images is challenging due to the wide variability of face appearances and the complexity of image backgrounds. This paper proposes a classification-based method for locating frontal faces in cluttered images. To improve the detection performance, we extract gradient direction features from local window images as the input of the underlying two-class classifier. The gradient direction representation provides better discrimination ability than the image intensity, and we show that the combination of gradient directionality and intensity outperforms the gradient feature alone. The underlying classifier is a polynomial neural network (PNN) on a reduced feature subspace learned by principal component analysis (PCA). The incorporation of the residual of subspace projection into the PNN was shown to improve the classification performance. The classifier is trained on samples of face and non-face images to discriminate between the two classes. The superior detection performance of the proposed method is justified in experiments on a large number of images. [Copyright &y& Elsevier]

• A new unified multi-modal method for text detection in Marathon and sports video. • It fuses gradient magnitude and direction coherence of text pixels in a new way. • Face and torso are detected by finding structural and spatial coherence. Detecting text located on the torsos of marathon runners and sports players in video is a challenging issue due to poor quality and adverse effects caused by flexible/colorful clothing, and different structures of human bodies or actions. This paper presents a new unified method for tackling the above challenges. The proposed method fuses gradient magnitude and direction coherence of text pixels in a new way for detecting candidate regions. Candidate regions are used for determining the number of temporal frame clusters obtained by K-means clustering on frame differences. This process in turn detects key frames. The proposed method explores Bayesian probability for skin portions using color values at both pixel and component levels of temporal frames, which provides fused images with skin components. Based on skin information, the proposed method then detects faces and torsos by finding structural and spatial coherences between them. We further propose adaptive pixels linking a deep learning model for text detection from torso regions. The proposed method is tested on our own dataset collected from marathon/sports video and three standard datasets, namely, RBNR, MMM and R-ID of marathon images, to evaluate the performance. In addition, the proposed method is also tested on the standard natural scene datasets, namely, CTW1500 and MS-COCO text datasets, to show the objectiveness of the proposed method. A comparative study with the state-of-the-art methods on bib number/text detection of different datasets shows that the proposed method outperforms the existing methods. [ABSTRACT FROM AUTHOR]