Search

Recent advance on highly realistic AI‐synthesised video approaches makes it hard to distinguish whether a speaker in video is real. Many existing DeepFakes detection approaches can be invalidated by using them as supervision in adversarial training, which in turn improves the performance of DeepFakes. This makes the problem a dilemma. However, human can recognise the identity of familiar persons by observing their motion styles. Inspired by this, the paper proposes a novel method to recognise original speaker's identity clues in DeepFakes videos by learning individual motion styles. As a biological signature, motion styles can neither be used in the training process of DeepFakes nor be modified to deceive detection methods without reducing realistic performance. Also, this paper proposes a novel pipeline to continuously restore the original frames from DeepFakes videos without knowing the DeepFakes approach in advance. Based on above ideas, our scheme makes it possible to restore DeepFakes videos. Through training the cross‐modal transfer module, the appearance embedding can be inferred from the identity code. Extensive experiments are conducted to reveal the effectiveness of the method, it shows for the first time that the original persons can be identified automatically and the original video from DeepFakes videos can be generated. [ABSTRACT FROM AUTHOR]

As one of the most fundamental tasks in computer vision, semantic segmentation assigns per pixel prediction of object categories. Training a robust model for semantic segmentation is challenging since pixel-level annotations are expensive to obtain. To alleviate the burden of annotations, the authors propose a weakly-supervised framework for zero-shot semantic segmentation, which can segment images having target classes without any pixel-level labelled instances. Under the assumption that the accessibility to image-level annotations of target classes does not violate the principle of zero pixel-level label in zero-shot semantic segmentation, we utilised image-level annotations to improve the proposed model's ability to extract pixel-level features. Furthermore, unlike existing zero-shot semantic segmentation methods, which use semantic embeddings as class embeddings to transfer knowledge from source classes to target classes, we use image-level features as their class embeddings to transfer knowledge since the distribution of pixel-level features is more similar to the distribution of image-level features rather than the distribution of semantic embeddings. Experimental results on the PASCAL-VOC data set under different data splits demonstrate that the proposed model achieves promising results. [ABSTRACT FROM AUTHOR]

Multi‐task cascaded convolutional neural network (MTCNN) is a human face detection architecture which uses a cascaded structure with three stages (P‐Net, R‐Net and O‐Net). The authors intend to reduce the computation time of the whole process of the MTCNN. They find that the non‐maximum suppression (NMS) processes after the P‐Net occupy over half of the computation time. Therefore, the authors propose a self‐fine‐tuning method which makes the control of computation time for the NMS process easier. Self‐fine‐tuning is a training trick which uses hard samples generated by P‐Net to retrain P‐Net. After self‐fine‐tuning, the distribution of human face probabilities generated by P‐Net is changed, and the tail of distribution becomes thinner. The control of the number of NMS input boxes can be made easier when the distribution has a thinner tail, and choosing a suitable threshold to filter the face boxes will generate less boxes. So the computation time can be reduced. In order to keep the performance of MTCNN, the authors still propose a landmark data set augmentation, which can enhance the performance of the self‐fine‐tuned MTCNN. From the experiments, it is found that the proposed scheme can significantly reduce the computation time of MTCNN. [ABSTRACT FROM AUTHOR]

Vector quantisation (VQ) is a widely used method for data compression or data clustering. This study proposes a fast codebook design method for image VQ. General particle swarm optimisation (PSO)‐K‐means hybrid methods for codebook design take too much time. To deal with this problem, the authors present a new clustering method, which takes a very short time and generates a pretty good codebook. This method contains four parts, i.e. pooling, PSO algorithm, reverse cast, and K‐means algorithm. Pooled images are used for the PSO process while original images are used for K‐means fine‐tuning, and these two processes connected by the reverse cast process. In the authors' experiment, this method can dramatically reduce the computation time by using big sizes of pooling window or enhance the codebook quality by using small sizes of pooling window. They can make the calculation time almost one‐tenth of that of the PSO‐K‐means hybrid method, and the scheme is also faster than the K‐means algorithm. Experimental results demonstrate that the main advantages of the proposed algorithm lie in the fact that it can reduce the computation time and enhance the quality of codebooks. [ABSTRACT FROM AUTHOR]

While promising progress has been achieved in the zero-shot learning (ZSL) task. The existing approaches still suffer from the strong bias problem between the unseen and seen classes. This Letter presents a unified feature generating framework equipped with a boundary decision loss to tackle this issue in ZSL. Specifically, the hybrid semantic and visual classification strategy is proposed, which can effectively align the bidirectional visual-semantic interactions. Furthermore, this Letter introduces a decision loss that optimises the decision boundary of seen and unseen classes to further alleviate the confusion of generated features. Extensive experiments on three popular datasets animals with attributes, Caltech-UCSD-Birds 200-2011, and SUN show that the proposed approach outperforms previous state-of-the-art works under both traditional ZSL and challenging generalised ZSL settings. [ABSTRACT FROM AUTHOR]

This study presents a hierarchical palmprint feature extraction and recognition approach based on multi‐wavelet and complex network (CN) since they can effectively decrease redundant information and enhance key points of main lines and wrinkles. The palmprint is first pre‐filtered and decomposed once using multi‐wavelet. Three components (LL1,2,3) corresponding to the pre‐filter except for diagonal component are extracted as the elementary features. Second, binary images (BLL1,2,3) are obtained by the average window method using different thresholds. Third, three series of dynamic evolution CN models (the 1st, 2nd, 3rd CNs) are constructed from global to local, which is based on the mosaiced images obtained from BLL1,2,3, BLL1 and four equally divided sub‐images of BLL1, respectively. Fourth, statistical features are extracted from complex networks, in which average degree and standard deviation of the degrees are extracted for the 1st CNs and average degrees are extracted for the 2nd and 3rd CNs. Fifth, the fisher feature is extracted using the linear discriminate analysis method. Finally, the nearest neighbourhood classifier is used to recognise palmprint. Based on the CASIA Palmprint Image Database, experimental results show that the proposed method can effectively recognise palmprint with good robustness and overcome the problem of small training samples number. [ABSTRACT FROM AUTHOR]

A novel approach to detect and localise anomalous events in crowed scenes by processing surveillance videos is introduced in this study. Unusual events are those that significantly differ from current dominated behaviours. The proposed approach both detects pixel‐level and block‐level anomalies. In pixel level, Gaussian mixture models are used to detect abnormalities. Block‐level detection segments the crowd into blocks according to pedestrian detection, and then anomalies are spotted and localised with a social force model. Experimental results using the USCD datasets Ped1 and Ped2 show that the proposed method performs favourably against state‐of‐the‐art methods. [ABSTRACT FROM AUTHOR]

Vector quantisation (VQ) shows a good performance for image coding with high‐compression ratios. However, there are many difficulties for image coding with VQ, especially the edge degradation and high‐computational complexity. To resolve these two problems, the authors propose a new coding method based on edge orientation patterns (EOPs) by classifying image blocks into nine classes according to their edge orientations. For colour image coding, 27 codebooks (nine for each colour component) are pre‐designed based on a series training images. In the encoding stage, an input colour image is decomposed into Y, Cb, and Cr components, and each component image is divided into non‐overlapping 4 × 4 blocks. For each block, eight edge orientation templates of size 4 × 4 are performed to determine its edge orientation. According to the edge orientation, each block is compressed by using the corresponding codebook. Essentially, the authors' scheme is a kind of classified VC (CVQ). Simulation results show that, their EOP‐based CVQ can largely improve the compression efficiency as well as speeding up the encoding process and it is sufficient to establish effectiveness of the authors' algorithm as compared with the existing techniques. [ABSTRACT FROM AUTHOR]

Recent advances in convolutional neural networks have achieved prominent results for object counting or locating in images. In these networks, some adopt detection‐based methods which require providing bounding‐box annotations for training and these methods usually cannot deal with the crowded cases well. Other networks adopt density regression‐based methods which require only point annotations for training but these methods can only predict the total count without providing any location information. In this Letter, the authors propose an attention guided convolutional neural network (AGNet) for both crowded object locating and counting using point annotation as supervision without annotated bounding boxes which are far more time‐consuming to label. Accordingly, they define a multi‐task objective loss function to lead the training of AGNet to achieve the goal of crowded object locating. The experimental results on four challenging benchmark datasets for crowded objects counting demonstrate the superiority of the proposed AGNet over recent state‐of‐the‐art approaches. [ABSTRACT FROM AUTHOR]

The intellectual property (IP) reuse‐based design model is vulnerable to misappropriation and fraud. The situation requires new IP protection mechanisms to guarantee the rights for both IP producers and IP users. In this study, the authors study the constraints used in circuit partitioning and hereby introduce two fingerprinting methods by cell perturbations. They evaluate the proposed methods on ISPD98 benchmark suits by the state‐of‐the‐art partitioner hMetis and cell manipulation. By tuning the fingerprinting strategies, the performance of fingerprinted designs is well maintained. Experimental result shows that the overhead of fingerprinting methods is trivial. The relative Hamming distance of cell fixing method is high, which indicates the method could yield distinct fingerprinting solutions. The relative Hamming distance of cell exchange method is low which is due to the low percentage of cell being fixed. [ABSTRACT FROM AUTHOR]

Recently, a large family of supervised or unsupervised manifold learning algorithms that stem from statistical or geometrical theory has been designed to solve the problem of pattern classification. In this study, consider the fact that the data are usually sampled from a low‐dimensional manifold space which resides in a high‐dimensional Euclidean space, the authors propose a novel two‐graph‐based supervised linear classification algorithm called neighbourhood sensitive preserving embedding (NSPE). Different from local linear embedding (LLE) (or neighbourhood preserving embedding (NPE)) which preserves the local neighbourhood structure with one graph, NSPE can discover both the intrinsic and discriminant structure of the data manifold by constructing two graphs, that is, the within‐class graph and the between‐class graph. Thus, the data are mapped into a subspace where the nearby points with the same label are close to each other, whereas the nearby points with different labels are far apart. As a classification method, besides being defined on training samples, NSPE is also defined on testing samples. Experiments carried on the real‐world face databases demonstrate that the results of all two‐graph‐based spectral methods are comparable and better than that of one‐graph‐based methods. [ABSTRACT FROM AUTHOR]

With the increasingly extensive application of networking technology, security of network becomes significant than ever before. Encryption algorithm plays a key role in construction of a secure network system. However, the encryption algorithm implemented on resource‐constrained device is difficult to achieve ideal performance. The issue of power consumption becomes essential to performance of data encryption algorithm. Many methods are proposed to evaluate the power consumption of encryption algorithms yet the authors do not ensure which one is effective. In this study, they give a comprehensive review for the methods of power evaluation. They then design a series of experiments to evaluate the effectiveness of three main types of methods by implementing several traditional symmetric encryption algorithms on a workstation. The experimental results show that external measurement and software profiling are more accurate than that of uninterruptible power system battery. The improvement of power consumption is 27.44 and 33.53% which implies the method of external measurement and software profiling is more effective in power consumption evaluation. [ABSTRACT FROM AUTHOR]

This Letter presents a new improved zero‐shot learning (ZSL) method based on meta‐learning that automatically adjusts metrics by multi‐layer perceptrons instead of using traditional Euclidean distance, cosine distance and other fixed metrics. In this way, the authors directly train the visual and semantic joint information of the images without adding additional auxiliary, which simply and effectively avoids the error derivation of unseen classes due to improper fixed metrics. Smooth one‐hot encoding rule is used to make the calculation of loss more reasonable. They also discuss the embedding methods to find a suitable low‐noise embedding space and analyse the model performance under the conventional setting and popular generalised ZSL setting. Experimental results show that the proposed automatic metrics method outperforms existing state‐of‐the‐art approaches in some parts of AwA2 and CUB datasets. [ABSTRACT FROM AUTHOR]

A new feature for image retrieval based on edge orientation patterns (EOPs) by classifying image blocks into nine classes according to their EOs. Each input colour image is decomposed into Y, Cb and Cr components, and each component image is divided into non‐overlapping 4 × 4 blocks. For each block, eight EO templates of size 4 × 4 are performed to determine its EO. On the other hand, vector quantisation (VQ) is an efficient clustering and classification technique for high‐dimensional spaces, and thus VQ is also used to extract the texture features. The retrieval simulation results show that the EOP‐histogram‐based features combined with the traditional VQ‐index‐histogram‐based features can largely improve the recall and precision performance. [ABSTRACT FROM AUTHOR]