Search

Facial data under non-rigid deformation are often assumed lying on a highly non-linear manifold. The conventional subspace clustering methods, such as Block Diagonal Representation (BDR), can only handle the high-dimensionality of facial data, ignoring the useful non-linear property embedded in data. Yet, discovering the local invariance in facial data remains a critical issue for face clustering. To this end, we propose a novel Block Diagonal Representation via Manifold learning (BDRM) in this paper. To be concrete, the manifold information within facial data can be learned by Locally Linear Embedding (LLE). Then manifold structure and block diagonal representation are considered jointly to uncover the intrinsic structure of facial data, which leads to a better representation for subsequent clustering task. Furthermore, the diffusion process is adopted to derive the final affinity matrix with context-sensitive, by which the learned affinity matrix can be spread and re-evaluated to enhance the connectivity of data belonging to the same intra-subspace. The extensive experimental results show that our proposed approach achieves a superior clustering performance against the state-of-the-art methods on both synthetic data and real-world facial data. [ABSTRACT FROM AUTHOR]

Robust face clustering enjoys a wide range of applications for gate passes, surveillance systems and security analysis in embedded sensors. Nevertheless, existing algorithms have limitations in finding accurate clusters when data contain noise (e.g., occluded face clustering and recognition). It is known that in subspace clustering, the ℓ 1 - and ℓ 2 -norm regularizers can improve subspace preservation and connectivity, respectively, and the elastic net regularizer (i.e., the mixture of the ℓ 1 - and ℓ 2 -norms) provides a balance between the two properties. However, existing deterministic methods have high per iteration computational complexities, making them inapplicable to large-scale problems. To address this issue, this paper proposes the first accelerated stochastic variance reduction gradient (RASVRG) algorithm for robust subspace clustering. We also introduce a new momentum acceleration technique for the RASVRG algorithm. As a result of the involvement of this momentum, the RASVRG algorithm achieves both the best oracle complexity and the fastest convergence rate, and it reaches higher efficiency in practice for both strongly convex and not strongly convex models. Various experimental results show that the RASVRG algorithm outperformed existing state-of-the-art methods with elastic net and ℓ 1 -norm regularizers in terms of accuracy in most cases. As demonstrated on real-world face datasets with different manually added levels of pixel corruption and occlusion situations, the RASVRG algorithm achieved much better performance in terms of accuracy and robustness. [ABSTRACT FROM AUTHOR]

Face clustering groups massive unlabeled face images according to their underlying identities and has proven to be a valuable tool for data analysis. Most recent studies have utilized graph convolutional networks (GCNs) to explore the structural properties of faces, thereby effectively achieving improved clustering performance. However, these methods usually suffer from computational intractability for large-scale graphs and tend to be sensitive to some postprocessing thresholds that serve to purify the clustering results. To address these issues, in this paper, we consider each pairwise relationship between two samples as a learning unit and infer clustering assignments by evaluating a group of pairwise connections. Specifically, we propose a novel clustering framework, named structure-enhanced pairwise feature learning (SEPFL), which mixes neighborhood information to adaptively produce pairwise representations for cluster identification. In addition, we design a combined density strategy to select representative pairs, thus ensuring training effectiveness and inference efficiency. The extensive experimental results show that SEPFL achieves better performance than other advanced face clustering techniques. [ABSTRACT FROM AUTHOR]

Abstract Face clustering groups massive unlabeled face images according to their underlying identities and has proven to be a valuable tool for data analysis. Most recent studies have utilized graph convolutional networks (GCNs) to explore the structural properties of faces, thereby effectively achieving improved clustering performance. However, these methods usually suffer from computational intractability for large-scale graphs and tend to be sensitive to some postprocessing thresholds that serve to purify the clustering results. To address these issues, in this paper, we consider each pairwise relationship between two samples as a learning unit and infer clustering assignments by evaluating a group of pairwise connections. Specifically, we propose a novel clustering framework, named structure-enhanced pairwise feature learning (SEPFL), which mixes neighborhood information to adaptively produce pairwise representations for cluster identification. In addition, we design a combined density strategy to select representative pairs, thus ensuring training effectiveness and inference efficiency. The extensive experimental results show that SEPFL achieves better performance than other advanced face clustering techniques.

When we are watching a video, often we may find it difficult to differentiate a character as we are unfamiliar with his/her face, especially if there are numerous actors/actresses or they are from different countries/cultures. There are also other circumstances like for deaf people or when people cannot hear the voice in noisy places(e.g. streets), a diarization method along with subtitles can be a more effective way to understand scripts. To address this, we proposed a video visualization system via face and speaker clustering. Given an input video, our system first separates the voice from the video and then extracts facial and voice features for face clustering and speaker clustering. Finally, the system finds the correspondence between face and speaker clustering results, and as a result, people could easily know when a character appears and who is the speaker in a video via our proposed video visualization system. [ABSTRACT FROM AUTHOR]

Face clustering is an effective method for taking advantage of unlabeled face data. Recent studies use graph convolutional networks (GCNs) to learn feature embeddings from the neighborhood information between face images. However, most of the face clustering methods require numerous overlapping subgraphs to characterize the local structure around the nodes, which causes significant redundancy. Moreover, the nonlinearity of the GCN itself increases the calculation complexity, which further reduces the model’s training efficiency. In this study, we propose a lightweight clustering framework, the confidence-based simple graph convolutional network (CSGCN), for face clustering, which achieves more accurate clustering results and significantly improves the efficiency of GCN-based face clustering. Specifically, CSGCN does not construct any subgraphs but convolves the entire graph as a whole and also removes the nonlinearity of the convolution in the graph convolution module, which further reduces the computational complexity. Subsequently, an effective new confidence score is constructed to better characterize the embedded features and to ensure that the subsequent clustering still maintains a high accuracy rate under the aforementioned model simplification. In addition, while most of the existing GCN-based methods are actually supervised, we construct an unsupervised confidence to make it more suitable for clustering tasks. Extensive experiments with MS-Celeb-1M, YouTube-Faces and DeepFashion datasets show that our method not only improves the clustering accuracy but also significantly reduces the execution time, whether in supervised or unsupervised models.

In this paper, a general optimization formulation is proposed for the subspace segmentation by low rank representation via the sparse-prompting quasi-rank function. We prove that, with the clean data from independent linear subspaces, the optimal solution to our optimization formulation not only is the lowest rank but also forms a block-diagonal matrix, which implies that it is reasonable to use any sparse-prompting quasi-rank function as the measure of the low rank in subspace clustering. With the data contaminated by Gaussian noise and/or gross errors, the alternating direction method of multipliers is applied to solving it and every sub-optimization problem has a closed-form optimal solution when the band restricted thresholding operator induced by its corresponding sparse-prompting function has an analytic expression, in which the gross errors part is replaced with the sparse-prompting matrix function. Finally, taking a specific sparse-prompting function, the fraction function, we conduct a series of simulations on different databases to get the performance of our algorithm tested, and experimental results show that our algorithm can obtain lower clustering error rate and higher value of evaluation indicators ACC, NMI and ARI than other state-of-the-art subspace clustering algorithms on different databases.

In this paper, we consider a problem of an automatic analysis of the emotional state of students during online classes based on video surveillance data. This problem is actual in the field of e-learning. We propose a novel neural network model for recognition of students' emotions based on video images of their faces and use it to construct an algorithm for classifying the individual and group emotions of students by video clips. At the first step, it performs detection of the faces and extracts their features followed by grouping the face of each student. To increase the accuracy, we propose to match students' names selected with the aid of the algorithms of the text recognition. At the second step, specially learned efficient neural networks perform the extraction of emotional features of each selected person, their aggregation with the aid of statistical functions, and the subsequent classification. At the final step, it is possible to visualize fragments of the video lesson with the most pronounced emotions of the student. Our experiments with some datasets from EmotiW (Emotion Recognition in the Wild) show that the accuracy of the developed algorithms is comparable with their known analogous. However, when classifying emotions, the computational performance of these algorithms is higher. [ABSTRACT FROM AUTHOR]

Mesh denoising is a classical task in mesh processing. Many state-of-the-art methods are still unable to quickly and robustly denoise multifarious noisy 3D meshes, especially in the case of high noise. Recently, neural network-based models have played a leading role in natural language, audio, image, video, and 3D model processing. Inspired by these works, we propose a data-driven mesh denoising method based on recurrent neural networks, which learns the relationship between the feature descriptors and the ground-truth normals. The recurrent neural network has a feedback loop before entering the output layer. By means of the self-feedback of neurons, the output of a recurrent neural network is related not only to the current input but also to the output of the previous moments. To deal with meshes with various geometric features, we use k-means to cluster the faces of the mesh according to geometric similarity and train neural networks for each category individually in the offline learning stage. Each network model, acting similar to a normal regression function, will map the geometric feature descriptor of each facet extracted from the mesh to the denoised facet normal. Then, the denoised normals are used to calculate the new feature descriptors, which become the input of the next similar regression model. In this system, three normal regression modules are cascaded to generate the last facet normals. Lastly, the model's vertex positions are updated according to the denoised normals. A large number of visual and numerical results have demonstrated that the proposed model outperforms the state-of-the-art methods in most cases. [ABSTRACT FROM AUTHOR]

Clustering of face images is a well-known clustering problem with several applications as for example the automated face tagging in photo albums. The performance of face clustering algorithms largely depends on the face representations that are used. Therefore, the exploitation of the internal face representation (embedding) obtained by feeding the face image to a deep face network has led to increased clustering performance. In this work, we perform face clustering by combining multiple representations of each image obtained from several deep face networks. The multiple deep representations are not exploited in a straightforward manner (simple vector concatenation). Instead a weight is assigned to each representation to reflect its importance on the clustering result and a weighted clustering algorithm is employed to automatically adjust the weights and, consequently, the contribution of each representation on the clustering solution. It should also be noted that in the proposed approach, the number of clusters (number of faces) is not given as input, but it is automatically estimated using the silhouette criterion. The conducted experiments have shown that the weighted combination of deep face representations leads to improved clustering performance compared to using single representations. [ABSTRACT FROM AUTHOR]

In an era of big data, face images captured in social media and forensic investigations, etc., generally lack labels, while the number of identities (clusters) may range from a few dozen to thousands. Therefore, it is of practical importance to cluster a large number of unlabeled face images into an efficient range of identities or even the exact identities, which can avoid image labeling by hand. Here, we propose adaptive facial imagery clustering that involves face representations, spectral clustering, and reinforcement learning (Q-learning). First, we use a deep convolutional neural network (DCNN) to generate face representations, and we adopt a spectral clustering model to construct a similarity matrix and achieve clustering partition. Then, we use an internal evaluation measure (the Davies–Bouldin index) to evaluate the clustering quality. Finally, we adopt Q-learning as the feedback module to build a dynamic multiparameter debugging process. The experimental results on the ORL Face Database show the effectiveness of our method in terms of an optimal number of clusters of 39, which is almost the actual number of 40 clusters; our method can achieve 99.2% clustering accuracy. Subsequent studies should focus on reducing the computational complexity of dealing with more face images. [ABSTRACT FROM AUTHOR]

Clustering is an important technique in data mining. It separates data points into different groups or clusters in such a way that objects in the same group are more similar to each other in some sense than with the objects in other groups. Gabor face clustering using affinity propagation and structural similarity index is composed of: A representation based on Gabor filters which has been shown to perform very well in face features, Affinity propagation clustering algorithm which is flexible, high speed, and does not require to specify the number of clusters, and structural similarity index which is a very powerful method for measuring the similarity between two images. Experimental results on two benchmark face datasets (LFW and IJB-B) show that our method outperforms well known clustering algorithms such as k-means, spectral clustering and Agglomerative. [ABSTRACT FROM AUTHOR]

With the rapid growth of online videos, it is crucial to generate overviews of videos to help audiences make viewing decisions and save time. Video summarization and video captioning are two of the most common solutions. In this paper, we proposed a new solution in the form of a series of scene-person pairs generated from our proposed video description scheme. This new formation takes substantially less time than watching video summaries and is more acceptable than video captions. In addition, our method can be generalized to different types of videos. We also proposed a face clustering method and a scene detection method. The experimental results indicate that our methods outperform other state-of-the-art methods and are highly generalizable. As an example, a demo application is developed to demonstrate the proposed description scheme. [ABSTRACT FROM AUTHOR]

An effective approach to automated movie content analysis involves building a network (graph) of its characters. Existing work usually builds a static character graph to summarize the content using metadata, scripts or manual annotations. We propose an unsupervised approach to building a dynamic character graph that captures the temporal evolution of character interaction. We refer to this as the character interaction graph (CIG). Our approach has two components: (i) an online face clustering algorithm that discovers the characters in the video stream as they appear, and (ii) simultaneous creation of a CIG using the temporal dynamics of the resulting clusters. We demonstrate the usefulness of the CIG for two movie analysis tasks: narrative structure (acts) segmentation and major character retrieval. Our evaluation on full-length movies containing more than 5000 face tracks shows that the proposed approach achieves superior performance for both the tasks. [ABSTRACT FROM AUTHOR]

Face clustering aims to group the face images without any label information into clusters, and has recently attracted considerable attention in machine learning and data mining. Many graph based clustering methods have been proposed and among which sparse representation (SR) and low-rank representation (LRR) are two representative methods for affinity graph construction. The clustering result may be inaccurate if the affinity graph is constructed with low quality. In this paper, we propose a novel face clustering method via learning a sparsity preserving low-rank graph (LSPLRG), where the initial affinity graph is derived on the sparse coefficients without any a priori graph or similarity matrix. In addition, an adaptive weighted matrix is imposed on the data reconstruction errors to enhance the role of important features, while a constraint on the representation matrix is to reduce the redundant features. By integrating the local distance regularization term into LRR, LSPLRG could exploit the global and local structures of data simultaneously. These appealing properties allow LSPLRG to well capture the intrinsic structure of data, and thus has potential to improve clustering performance. Experiments conducted on several face image databases demonstrate the effectiveness and robustness of LSPLRG compared with several state-of-the-art subspace clustering methods. [ABSTRACT FROM AUTHOR]

In this paper, we propose an unconstrained face verification approach that is dependent on Hybrid Siamese architecture under limited resources. The general face verification trend suggests that larger training datasets and/or complex architectures lead to higher accuracy. The proposed approach tends to achieve high accuracy while using a small dataset and a simple architecture by directly learn face's similarity/dissimilarity from raw face pixels, which is critical for various applications. The proposed architecture has two branches; the first part of these branches is trained independently, while the other parts shared their parameters. A multi-batch algorithm is utilized for training. The training process takes a few hours on a single GPU. The proposed approach achieves near-human accuracy (98.9%) on the Labeled Faces in theWild (LFW) benchmark, which is competitive with other techniques that are presented in the literature. It reaches 99.1% on the Arabian faces dataset. Moreover, features learned by the proposed architecture are used in building a face clustering system that is based on an updated version of the Density-Based Spatial Clustering of Applications with Noise (DBSCAN). To handle the cluster quality challenge, a novel post-clustering optimization procedure is proposed. It outperforms popular clustering approaches, like K-Means and spectral by 0.098 and up to 0.344 according to F1-measure. [ABSTRACT FROM AUTHOR]

Despite great progress in face recognition tasks achieved by deep convolution neural networks (CNNs), these models often face challenges in real world tasks where training images gathered from Internet are different from test images because of different lighting condition, pose and image quality. These factors increase domain discrepancy between training (source domain) and testing (target domain) database and make the learnt models degenerate in application. Meanwhile, due to lack of labeled target data, directly fine-tuning the pre-learnt models becomes intractable and impractical. In this paper, we propose a new clustering-based domain adaptation method designed for face recognition task in which the source and target domain do not share any classes. Our method effectively learns the discriminative target feature by aligning the feature domain globally, and, at the meantime, distinguishing the target clusters locally. Specifically, it first learns a more reliable representation for clustering by minimizing global domain discrepancy to reduce domain gaps, and then applies simplified spectral clustering method to generate pseudo-labels in the domain-invariant feature space, and finally learns discriminative target representation. Comprehensive experiments on widely-used GBU, IJB-A/B/C and RFW databases clearly demonstrate the effectiveness of our newly proposed approach. State-of-the-art performance of GBU data set is achieved by only unsupervised adaptation from the target training data. [ABSTRACT FROM AUTHOR]

Nonsmooth nonnegative matrix factorization (nsNMF) is capable of producing more localized, less overlapped feature representations than other variants of NMF while keeping satisfactory fit to data. However, nsNMF as well as other existing NMF methods are incompetent to learn hierarchical features of complex data due to its shallow structure. To fill this gap, we propose a deep nsNMF method coined by the fact that it possesses a deeper architecture compared with standard nsNMF. The deep nsNMF not only gives part-based features due to the nonnegativity constraints but also creates higher level, more abstract features by combing lower level ones. The in-depth description of how deep architecture can help to efficiently discover abstract features in dnsNMF is presented, suggesting that the proposed model inherits the major advantages from both deep learning and NMF. Extensive experiments demonstrate the standout performance of the proposed method in clustering analysis.

In the normalized cut (Ncut) process, it is crucial to construct an appropriate affinity matrix. The affinity matrix is generally limited to pairwise similarity relations. However, in practice, it is necessary to use high-order affinities in several computer vision applications such as motion segmentation. In this paper, by using high-order singular value decomposition techniques, we derive a high-order affinity model directly from the Ncut relaxation formula, called high-order normalized cut (HNcut). However, in practice, it cannot directly utilize the high-order affinity matrix because of the computational resources required. To address this issue, we adopt and improve various techniques to make the proposed method more practical such as sampling strategy. Finally, we analyze the upper error bound of our algorithm based on matrix perturbation theory. To demonstrate the performance of our HNcut, we compare it with some existing algorithms for the motion segmentation and face clustering problems.

This paper is focused on the automatic extraction of persons and their attributes (gender, year of born) from album of photos and videos. A two-stage approach is proposed in which, firstly, the convolutional neural network simultaneously predicts age/gender from all photos and additionally extracts facial representations suitable for face identification. Here the MobileNet is modified and is preliminarily trained to perform face recognition in order to additionally recognize age and gender. The age is estimated as the expected value of top predictions in the neural network. In the second stage of the proposed approach, extracted faces are grouped using hierarchical agglomerative clustering techniques. The birth year and gender of a person in each cluster are estimated using aggregation of predictions for individual photos. The proposed approach is implemented in an Android mobile application. It is experimentally demonstrated that the quality of facial clustering for the developed network is competitive with the state-of-the-art results achieved by deep neural networks, though implementation of the proposed approach is much computationally cheaper. Moreover, this approach is characterized by more accurate age/gender recognition when compared to the publicly available models.

In an era of big data, face images captured in social media and forensic investigations, etc., generally lack labels, while the number of identities (clusters) may range from a few dozen to thousands. Therefore, it is of practical importance to cluster a large number of unlabeled face images into an efficient range of identities or even the exact identities, which can avoid image labeling by hand. Here, we propose adaptive facial imagery clustering that involves face representations, spectral clustering, and reinforcement learning (Q-learning). First, we use a deep convolutional neural network (DCNN) to generate face representations, and we adopt a spectral clustering model to construct a similarity matrix and achieve clustering partition. Then, we use an internal evaluation measure (the Davies–Bouldin index) to evaluate the clustering quality. Finally, we adopt Q-learning as the feedback module to build a dynamic multiparameter debugging process. The experimental results on the ORL Face Database show the effectiveness of our method in terms of an optimal number of clusters of 39, which is almost the actual number of 40 clusters; our method can achieve 99.2% clustering accuracy. Subsequent studies should focus on reducing the computational complexity of dealing with more face images.

In order to solve the problem that sparse subspace clustering cannot effectively cluster the dataset under non-independent assumption, this paper proposes the sparse subspace clustering with low-rank transformation, which merges the low-rank transformation into sparse subspace clustering. The low-rank transformation can not only reduce variations within the subspaces, but also increase separations between the subspaces. This guarantees that property of these datasets is altered from inconsistent independent assumption to asymptotic consistent independent assumption, ultimately to consistent independent assumption. Sparse subspace clustering with low-rank transformation is formulated as the minimization optimizations with two variables, which can be solved by an iterative strategy including the clustering and computing the optimal transformation. Experimental results on the synthetic datasets show that sparse subspace clustering with low-rank transformation is robust to synthetic dataset with noise and its classification error is approximately 75% lower than that of sparse subspace clustering. Experimental results on face clustering on the Yale Extend B dataset and motion segmentation on Hopkins 155 dataset show that the proposed approach significantly outperforms sparse subspace clustering. [ABSTRACT FROM AUTHOR]

• In this article, we focus on the Sparse Subspace Clustering (SSC) method. • We present a novel method based on a mixture of Riemannian Kernels. • The proposed method succeeds to improve clustering results compared to other state of the art methods. • The experimental context is essentially the image and video Face clustering. Recently, clustering in the Riemannian manifolds has received a great interest. The main specificity of this kind of spaces is their ability to detect nonlinear forms in the real world data groups. Accordingly, numerous Euclidean clustering algorithms have been emerged to the Riemannian framework. One of them is the Sparse Subspace Clustering (SSC) method. Within this context, an interesting method named Kernel Sparse Subspace Clustering in the Riemannian manifold (KSSCR) has been proposed. In order to improve the clustering performance of the SSC method, KSSCR is based on the idea of data projection into a Reproducing Kernel Hilbert Space (RKHS) through a Riemannian kernel. This modification in terms of projection space has yielded to significant clustering results compared to the original SSC. This paper proposed a Multi-Kernel SSCR (MKSSCR) method. The idea is to create a linear combination of a set of Riemannian kernels. In fact, by using a single kernel, the faraway points will be neglected. Thus, the key motivation of this work is to use a mixture of kernels. The purpose here is to emphasis the closest data points without ignoring the faraway ones in the process of selecting neighbors for each data point. Several experiments carried out on varied face clustering data sets demonstrate the clustering accuracy improvement by the proposed method compared to other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Face clustering, a technique for automatically annotating large-scale face data, has made significant advancements with the advent of graph convolutional networks (GCNs). Despite their success, GCNs can suffer from decreased performance due to conflicting information passed along noisy edges of a graph. To address this issue, we propose a novel framework named progressive structure enhancement GCN (PSE-GCN), which combines graph structure learning with graph-guided feature aggregation. Our PSE-GCN framework includes a dynamic graph construction (DGC) module that enhances local relationships and suppresses global noise, thereby improving the quality of the graph. By stacking multiple DGCs, PSE-GCN progressively refines the graph quality and yields discriminative features for various clustering tasks. Additionally, we introduce a subgraph-based neighborhood re-ranking (SNR) mechanism that improves graph homogeneity by rearranging the candidate neighbors of each face based on structural similarity at the subgraph level. Our experimental results, conducted on several popular benchmarks, not only demonstrate the effectiveness of PSE-GCN, but also show that it outperforms state-of-the-art methods, e.g., 93.50% in pairwise F-score on the MS-Celeb-1M dataset. [ABSTRACT FROM AUTHOR]

In this paper, we propose an unconstrained face verification approach that is dependent on Hybrid Siamese architecture under limited resources. The general face verification trend suggests that larger training datasets and/or complex architectures lead to higher accuracy. The proposed approach tends to achieve high accuracy while using a small dataset and a simple architecture by directly learn face’s similarity/dissimilarity from raw face pixels, which is critical for various applications. The proposed architecture has two branches; the first part of these branches is trained independently, while the other parts shared their parameters. A multi-batch algorithm is utilized for training. The training process takes a few hours on a single GPU. The proposed approach achieves near-human accuracy (98.9%) on the Labeled Faces in the Wild (LFW) benchmark, which is competitive with other techniques that are presented in the literature. It reaches 99.1% on the Arabian faces dataset. Moreover, features learned by the proposed architecture are used in building a face clustering system that is based on an updated version of the Density-Based Spatial Clustering of Applications with Noise (DBSCAN). To handle the cluster quality challenge, a novel post-clustering optimization procedure is proposed. It outperforms popular clustering approaches, like K-Means and spectral by 0.098 and up to 0.344 according to F1-measure.

YouTube is the most popular platform for streaming of user-generated videos. Nowadays, professional YouTubers are organized in so-called multichannel networks (MCNs). These networks offer services such as brand deals, equipment, and strategic advice in exchange for a share of the YouTubers' revenues. A dominant strategy to gain more subscribers and, hence, revenue is collaborating with other YouTubers. Yet, collaborations on YouTube have not been studied in a detailed quantitative manner. To close this gap, first, we collect a YouTube dataset covering video statistics over 3 months for 7,942 channels. Second, we design a framework for collaboration detection given a previously unknown number of persons featured in YouTube videos. We denote this framework, for the detection and analysis of collaborations in YouTube videos using a Deep Neural Network (DNN)-based approach, as CATANA. Third, we analyze about 2.4 years of video content and use CATANA to answer research questions guiding YouTubers and MCNs for efficient collaboration strategies. Thereby, we focus on (1) collaboration frequency and partner selectivity, (2) the influence of MCNs on channel collaborations, (3) collaborating channel types, and (4) the impact of collaborations on video and channel popularity. Our results show that collaborations are in many cases significantly beneficial regarding viewers and newly attracted subscribers for both collaborating channels, often showing more than 100% popularity growth compared with noncollaboration videos. [ABSTRACT FROM AUTHOR]

Collaborative representation methods, such as sparse subspace clustering (SSC) and sparse representation-based classification (SRC), have achieved great success in face clustering and classification by directly utilizing the training images as the dictionary bases. In this paper, we reveal that the superior performance of collaborative representation relies heavily on the sufficiently large class separability of the controlled face datasets such as Extended Yale B. On the uncontrolled or undersampled dataset, however, collaborative representation suffers from the misleading coefficients of the incorrect classes. To address this limitation, inspired by the success of linear discriminant analysis (LDA), we develop a superposed linear representation classifier (SLRC) to cast the recognition problem by representing the test image in term of a superposition of the class centroids and the shared intra-class differences. In spite of its simplicity and approximation, the SLRC largely improves the generalization ability of collaborative representation, and competes well with more sophisticated dictionary learning techniques, on the experiments of AR and FRGC databases. Enforced with the sparsity constraint, SLRC achieves the state-of-the-art performance on FERET database using single sample per person. [ABSTRACT FROM AUTHOR]

In this paper, we simulate the learning way of human to propose a self-learning framework for face clustering. Specifically, we first perform a decorrelation operation on face images through patch-based two-dimensional reconstruction, which has a similar function to the retina. Then we group the semantically similar faces by using a novel self-paced learning model, which is inspired by three major observations: (i) The learning process of human gradually proceeds from easy to complex tasks; (ii) The prior knowledge of human might change with the increase of learned experience; (iii) More prior knowledge usually leads to better prediction accuracy. Experiments on benchmark face databases demonstrate the effectiveness and efficiency of the proposed framework. [ABSTRACT FROM AUTHOR]

Clustering face images according to their latent identity has two important applications: 1) grouping a collection of face images when no external labels are associated with images, and 2) indexing for efficient large scale face retrieval. The clustering problem is composed of two key parts: representation and similarity metric for face images, and choice of the partition algorithm. We first propose a representation based on ResNet, which has been shown to perform very well in image classification problems. Given this representation, we design a clustering algorithm, Conditional Pairwise Clustering (ConPaC), which directly estimates the adjacency matrix only based on the similarities between face images. This allows a dynamic selection of number of clusters and retains pairwise similarities between faces. ConPaC formulates the clustering problem as a Conditional Random Field model and uses Loopy Belief Propagation to find an approximate solution for maximizing the posterior probability of the adjacency matrix. Experimental results on two benchmark face datasets (LFW and IJB-B) show that ConPaC outperforms well known clustering algorithms such as k -means, spectral clustering, and approximate Rank-order. Additionally, our algorithm can naturally incorporate pairwise constraints to work in a semi-supervised way that leads to improved clustering performance. We also propose a k -NN variant of ConPaC, which has a linear time complexity given a k -NN graph, suitable for large datasets. [ABSTRACT FROM PUBLISHER]

• Face clustering is first reformulated as non-overlapping community detection. • Three new metrics for facial image clustering are introduced. • A novel framework achieves new state-of-the-arts in face clustering. [Display omitted] Face clustering is a critical task in computer vision due to the increasing number of applications such as augmented reality or photo album management. The primary challenge in this task arises from the imperfections in image feature representations. Given image features extracted from an existing pre-trained representation model, it remains an unresolved problem that how to leverage the inherent characteristics of similarities among unlabelled images to improve the clustering performance. In order to solve face clustering in an unsupervised manner, we develop an effective and robust framework named as Adapt-Infomap. First, we reformulate face clustering as a process of non-overlapping community detection. Specially, Adapt-Infomap achieves face clustering by minimizing the entropy of information flows (also known as the map equation) on an affinity graph of images. Since the affinity graph of images might contain noisy edges, we develop an outlier detection strategy in Adapt-Infomap to adaptively refine the affinity graph. Experiments with ablation studies demonstrate that Adapt-Infomap significantly outperforms existing methods and achieves new state-of-the-arts on three popular large-scale datasets for face clustering, e.g. , an absolute improvement of more than 10 % and 3 % comparing with prior unsupervised and supervised methods respectively in terms of average of Pairwise F-score. [ABSTRACT FROM AUTHOR]

Given a large collection of unlabeled face images, we address the problem of clustering faces into an unknown number of identities. This problem is of interest in social media, law enforcement, and other applications, where the number of faces can be of the order of hundreds of million, while the number of identities (clusters) can range from a few thousand to millions. To address the challenges of run-time complexity and cluster quality, we present an approximate Rank-Order clustering algorithm that performs better than popular clustering algorithms (k-Means and Spectral). Our experiments include clustering up to 123 million face images into over 10 million clusters. Clustering results are analyzed in terms of external (known face labels) and internal (unknown face labels) quality measures, and run-time. Our algorithm achieves an F-measure of 0.87 on the LFW benchmark (13 K faces of 5,749 individuals), which drops to 0.27 on the largest dataset considered (13 K faces in LFW + 123M distractor images). Additionally, we show that frames in the YouTube benchmark can be clustered with an F-measure of 0.71. An internal per-cluster quality measure is developed to rank individual clusters for manual exploration of high quality clusters that are compact and isolated. [ABSTRACT FROM PUBLISHER]