The covariance descriptor which is a symmetric positive definite (SPD) matrix, has recently attracted considerable attentions in computer vision. However, it is not trivial issue to handle its non-linearity in semi-supervised learning. To this end, in this paper, a semi-supervised sparse subspace clustering on SPD manifolds is proposed, via considering the intrinsic geometric structure within the manifold-valued data. Experimental results on two databases show that our method can provide better clustering solutions than the state-of-the-art approaches thanks to incorporating Riemannian geometry structure.
With the fast development of social networks, high-dimensionality is becoming an intractable problem in many machine learning and computer vision tasks. This phenomenon also exists in the field of multi-label classification. So far many supervised or semi-supervised multi-label feature selection methods have been proposed to reduce the feature dimension of training samples. However, almost all existing feature selection works focus on multi-label learning with complete labels. In fact, labels are very expensive to obtain and the training instances usually have an incomplete/partial set of labels (some labels are randomly missed). Very few researchers pay attention to the problem of multi-label feature selection with missing labels. In this paper, we propose a robust model to solve the above problem. We recover the missing labels by a linear regression model and select the most discriminative feature subsets simultaneously. The effective $$l_{2,p}$$ -norm $$\left( {0 < p \le 1} \right) $$ regularization is imposed on the feature selection matrix. The iterative reweighted least squares (IRLS) algorithm is used to solve the optimization problem. To verify the effectiveness of the proposed method, we conduct experiments on five benchmark datasets. Experimental results show that our method has superior performance over the state-of-the-art algorithms.
This paper considers the problem of clustering uncertain objects whose locations are uncertain and described by probability density functions (pdf). Though K-means has been extended to UK-means for handling uncertain data, most existing works only focus on improving the efficiency of UK-means. However, the clustering quality of UK-means is rarely considered in existing works. The weights of objects are assumed same in existing works. However, the weights of objects which are far from their cluster representatives should not be the same as the weights of objects which are close to their cluster representatives. Thus, we propose an AdaUK-means to group the uncertain objects by considering the weights of objects in this article. In AdaUK-means, the weights of objects will be adjusted based on the correlation between objects by using Adaboost. If the object pairs are must-link but grouped into different clusters, the weights of the objects will be increased. In our ensemble model, AdaUK-means is run several times, then the objects are assigned by a voting process. Finally, we demonstrate that AdaUK-means performs better than UK-means on both synthetic and real data sets by extensive experiments.
A novel semantic segmentation method is proposed, which consists of the following three parts: (I) First, a simple yet effective data augmentation method is introduced without any extra GPU memory cost during training. (II) Second, a deeper residual network is constructed through three effective techniques: dilated convolution, LSTM network and multi-scale prediction. (III) Third, an online hard pixels mining is adopted to improve the segmentation performance. We combine these three parts to train an end-to-end network and achieve a new state-of-the-art segmentation accuracy of 79.3% on PASCAL VOC 2012 test set at the time of submission.
Structure information is a hot spot currently in the domain of computer vision. As many people had applied structure information to their method, few people employed degradation information to their algorithm. However the degradation itself contains some important information. In this paper, we introduce a Structured Degradation Model with degradation assessment of the target to solve the tracking problem. To track the target in non-uniform degraded video, autocorrelation is used to generate the direction map and Tenengrad is used to extract the degradation degree of each target part. In our Structured Degradation Model, an undirected graph of the target is generated to track the target. The nodes of the graph are the target parts and the edges are the interactions between the parts. Experimental result shows that our method performs well especially for object tracking in degraded video.
License plate detection is a crucial part in license plate recognition systems and is often considered as a solved problem. However, there are still plenty of complex scenes where the current methods are invalidated. In order to increase the performance in these scenes, we propose a novel character-based method to detect multiple license plates in complex images. Firstly, a preprocessing step is performed. Then we use a modified maximally stable extremal region (MSER) based detector called MSER-+ to detect the possible character regions. Some of the regions are removed according to their geographical information. Hierarchical morphology helps to connect candidate MSERs of various sizes. The regions satisfying some geographical limits will be fed into a convolutional neural network (CNN) model for further verification. Extensive experimental results validate that our method works well in a large variety of complex scenes.
Touchless fingerprint recognition with high acceptance, high security, hygiene advantages, is currently a hot research field of biometrics, but because of the different image principle of the non-contact fingerprint image and contact fingerprint image, the difference of the two fingerprint image is large. There are still a small number of fuzzy regions in the non-contact fingerprint image after pretreatment, and the traditional method of extracting the future from the detail points can lead to a serious decline in recognition accuracy because of false points. In this paper, the non-contact pretreatment in our laboratory is used according to the characteristics of the contactless fingerprint image, the LBP operator, LGC operator and their improve algorithms are used for image processing; the nearest neighbor classifier is used for feature matching. The experimental result shows that the contactless fingerprint feature extraction method proposed in this paper can obtain higher division fingerprint feature.
This paper investigates on the deep learning approaches for the social touch gesture recognition. Several types of neural network architectures are studied with a comprehensive experiment design. First, recurrent neural network using long short-term memory (LSTM) is adopted for modeling the gesture sequence. However, for both handcrafted features using geometric moment and feature extraction using convolutional neural network (CNN), LSTM cannot achieve satisfactory performances. Therefore, we propose to use the 3D CNN to model a fixed length of touch gesture sequence. Experimental results show that the 3D CNN approach can achieve a recognition accuracy of 76.1 % on the human-animal affective robot touch (HAART) database in the recognition of social touch gestures challenge 2015, which significantly outperforms the best submitted system of the challenge with a recognition accuracy of 70.9 %.
Face detection in unconstrained environment is a challenge problem. Recent studies show that deep convolutional networks (DCNs) have achieved outstanding performance on this task, but most of them have multiple stages (e.g., region proposal, classification), which are complex and time-consuming in practice. In this paper, we propose a fully convolutional network (FCN) framework which can be trained straightforward in an end-to-end manner. In our network, hierarchical feature layers with different resolutions are used to detect different scale faces. For each hierarchical layer, a specific default boxes set with different aspect ratios and scales is associated with each map cell. At prediction time, the network generates confidence scores for the default boxes and produces offsets of default boxes to get better bounding boxes of faces. The predictions of each hierarchical layer are combined into final detection result. Experimental results on the AFW and FDDB datasets confirm the effectiveness of our method.
Many dimensionality reduction problems can be formulated as a trace ratio form, i.e. argmax_𝐖Tr(𝐖^T 𝐒_p 𝐖) / Tr(𝐖^T 𝐒_t 𝐖) , where 𝐒_p and 𝐒_t represent the (dis)similarity between data, 𝐖 is the projection matrix, and Tr(· ) is the trace of a matrix. Some representative algorithms of this category include principal component analysis (PCA), linear discriminant analysis (LDA) and marginal Fisher analysis (MFA). Previous research focuses on how to solve the trace ratio problems with either (generalized) eigenvalue decomposition or iterative algorithms. In this paper, we analyze an algorithm that transforms the trace ratio problems into a series of trace difference problems, i.e. argmax_𝐖Tr[(𝐖^T (𝐒_p - λ𝐒_t )𝐖] , and propose the necessary and sufficient conditions for the existence of the optimal solution of trace ratio problems. The correctness of this theoretical result is proved. To evaluate the applied algorithm, we tested it on three face recognition applications. Experimental results demonstrate its convergence and effectiveness.
It is well-known that model combination can improve prediction performance of regression model. We investigate the model combination of Lasso with regularization path in this paper. We first define the prediction risk of Lasso estimator, and prove that Lasso regularization path contains at least one prediction consistent estimator. Then we establish the prediction consistency for convex combination of Lasso estimators, which gives the mathematical justification for model combination of Lasso on regularization path. With the inherent piecewise linearity of Lasso regularization path, we construct the initial candidate model set, then select the models for combination with Occam’s Window method. Finally, we carry out the combination on the selected models using the Bayesian model averaging. Theoretical analysis and experimental results suggest the feasibility of the proposed method.
Position information of sound source provides important cues for many audio analysis tasks. In this paper, we present a simple yet effective simultaneous audio source location and microphone placement approach to obtain the position information of stationary sound source. Motivation by the fact that audio source location and microphone placement can help each other, we consider these tasks as a joint inference framework so that more contextual information can be exploited. By fusing geometric properties of the sliding microphone-pair system with an error theory analysis, our location approach can achieve higher accuracy than conventional methods theoretically. Moreover, experimental results on our built real sound source location system are presented to demonstrate the effectiveness of the proposed approach.
Emotion recognition is a significant research filed of pattern recognition and artificial intelligence. The Multimodal Emotion Recognition Challenge (MEC) is a part of the 2016 Chinese Conference on Pattern Recognition (CCPR). The goal of this competition is to compare multimedia processing and machine learning methods for multimodal emotion recognition. The challenge also aims to provide a common benchmark data set, to bring together the audio and video emotion recognition communities, and to promote the research in multimodal emotion recognition. The data used in this challenge is the Chinese Natural Audio-Visual Emotion Database (CHEAVD), which is selected from Chinese movies and TV programs. The discrete emotion labels are annotated by four experienced assistants. Three sub-challenges are defined: audio, video and multimodal emotion recognition. This paper introduces the baseline audio, visual features, and the recognition results by Random Forests.
This paper focuses on a data cleaning method to denoise and detect outliers of earthen site monitoring data with wireless sensor network (WSN). A data cleaning method, named DC_ESVS is proposed, which is based on the temporal and spatial characteristics of monitoring data with WSN. Using the cubic exponential smoothing algorithm and voting strategy, it can denoise and detect outliers of earthen site monitoring data based on the decision rule. We conduct various experiments on the dataset of the monitoring data of Xi’an Tang Hanguangmen city wall site with WSN to show detection accuracy of the presented method. Experimental results on anther dataset of the monitoring data of the Ming Great Wall in Shaanxi also show good performance of the proposed method.
Finding discriminative feature is crucial for building a high-performance object detection system, which has an effect on the detection speed and accuracy. In this paper, we propose a novel discriminative weighted pooling feature based on the multiple channel maps for multi-view face detection. The color and shape statistics of face structure can be utilized to enhance the discriminative ability of the box filter, which is generalized from the square channel filter. The discriminative information can be obtained with LDA and imbalance embedding LDA method, which is superior to the baseline box filter. The experimental result on the FDDB dataset shows that our proposed method has some advantages in accuracy or speed when compared with many other state-of-the-art methods.
Community structure is one of the most important properties of complex networks and a keypoint to understanding and exploring real-world networks. One popular technique for community detection is matrix-based algorithms. However, existing matrix-based community detection models, such as nonnegative matrix factorization, spectral clustering and their variants, fit the data in a Euclidean space and have ignored the local consistency information which is crucial when discovering communities. In this paper, we propose a novel framework of latent space clustering to cope with community detection, by incorporating the clique-based locally consistency into the original objective functions to penalize the latent space dissimilarity of the nodes within the clique. We evaluate the proposed methods on both synthetic and real-world networks and experimental results show that our approaches significantly improve the accuracy of community detection and outperform state-of-the-art methods, especially on networks with unclear structures.
Many similar shaped scripts are used all over the world today. Scripts identification with similar shaped characters is one of the difficulties in script identification field and it need to be resolved. However, there are a little report about identification of Central Asian countries and Chinese Minority scripts, which identification of similar scripts. In this paper, a multi-script database was established, which are including 2200 plain document images with different resolution in 11 scripts such as English, Chinese, Arabic, Russian, Uyghur, Mongol, Tibet, Turkish, Kyrgyzstani, Uzbekistani and Tajikistani. Then, HSV features were extracted from each whole page image and they were classified by using BP neural network classifier. After experiment in our system, it is achieved 88.14 % of average identification rate and 99.0 % of highest identification rate in our experiment with the dataset. Experimental results indicated that HSV features were effective feature for identify these scripts.
Dump truck recognition plays an important role in the state-owned land surveillance system, which aims at fore-warning illegal construction. However, there is no special algorithm for dump truck recognition. In this paper, we explore a dump truck recognition algorithm combing structure components projection with spatial relationship (SCPSR). Instead of detecting dump truck directly as a whole, we propose a dump truck recognition algorithm based on foreground detection and components detection. An improved three frames difference method is used for foreground detection. Inspired by structure feature of dump truck components, we first locate the wheels by its valley feature on gray-scale image, and then search the candidate cab and hopper zones with the help of spatial relationship. Further, cab and hopper zones are determined by the components projection. Combining foreground detection with components detection method, the system is able to provide real-time and reliable vehicle supervision results. Experiments on real site videos demonstrate promising performance of the proposed algorithm.
Disease gene identification is of great significance for the treatment of genetic disorders. In recent years, the rapid development of high-throughput sequencing technologies has brought great revolution for disease gene identification methods. Network-based methods are now the most efficient component for disease gene identification, while the most of current methods pay only attention to the local topological attributes regardless of the global distribution. In this paper, we proposed to apply the random walk algorithm to extract global features for each gene and finally used binary logistic regression model to identify whether a gene belongs to the given disease. We also integrate the local features and global features into a complex feature vector to improve the identification performance. The experimental results show that the global feature is of great efficiency for disease gene identification. We organize the global feature into different kinds of feature vectors and we can get higher AUC scores than other state-of-the-art methods for all these feature vectors.
