Search

Most of the existing blind omnidirectional image quality assessment (BOIQA) methods are based on data-driven approach where the end-to-end neural network or deep learning tools are mainly used for feature extraction. However, it usually lacks interpretability and is difficult to discover the perceptual mechanism behind. In this paper, from the perspective of perception modeling, we propose a novel multi-angle projection based BOIQA (MP-BOIQA) method. Considering the omnibearing and near eye display characteristics with head mounted display, multiple color cubemap projection images with respect to different viewpoints are grouped as the color omnidirectional distortion (COD) units so as to simulate the user’s viewing behavior in subjective quality assessment. In the designed multi-angle projection based feature extractor, tensor decomposition is implemented on each COD unit for dimensionality reduction, and piecewise exponential fitting is used to get the distribution of mean subtracted contrast normalized coefficients of the unit’s feature matrices in tensor domain. Finally, the extracted features are pooled with random forest. The experimental results on three omnidirectional image quality datasets show that the MP-BOIQA method can deliver highly competitive performance compared with some representative full-reference quality assessment methods, as well as some state-of-the-art BOIQA methods. [ABSTRACT FROM AUTHOR]

This paper proposes a novel facial landmark detection (FLD) algorithm for use in real driving situations. The proposed algorithm is based on an ensemble of local weighted random forest regressor (WRFR) with random sampling consensus (RANSAC) and explicit global shape models, and considers the dynamic and irregular characteristics of driving. In this paper, to estimate the offset distance from a landmark and a reference point, we first detect the nose region as a reference point. Next, we propose a local WRFR to maintain the generality with a small number of regression trees. With the WRFR, we adopt RANSAC instead of the averaging or median of offsets to handle the problem of sensitivity to outlier offsets, and we estimate the accurate 2D offset vector. To identify the erroneous positions of local landmarks and rearrange the overall landmark layout, we adopt the global face models based on the spatial relation between landmarks. Using the unified framework of the proposed FLD, our proposed algorithm is robust to large head poses and partial occlusions caused by a driver’s hair or sunglasses. For a benchmark data set considering real driving situations, we construct a data set called a face alignment data set used in driving (FADID) using a near-infrared camera for FLD under real driving situations. We apply the proposed algorithm to various driving sequences in FADID, and the results show that its FLD performance is better than that of other state-of-the-art methods, while the computational speed is high for real-time applications such as driver-state monitoring systems. [ABSTRACT FROM AUTHOR]

With the development of video surveillance in public safety field, there is an increasing research on person re-identification (re-id). In this paper, we address the image-to-video person re-id, in which the probe is an image and the gallery is consists of videos captured by nonoverlapping cameras. Compared with image, video sequence contains more temporal information that can be explored to improve the performance of re-identification system. However, it is challenging to model temporal information in the matching process of image-to-video person re-id. In this paper, we proposed a novel temporally memorized similarity learning neural network for this problem. In specific, the proposed network mainly consisted of two parts, including feature representation sub-network and similarity sub-network. In the first part, we adopted a convolutional neural network (CNN) to extract features from the input image. Given a video sequence of a person, features were first extracted from each its frame by using CNN and further forward to a long shot term memory (LSTM) network to encode the temporal information of video sequence. The outputs of LSTM were concatenated together as the feature vector of video sequences. Finally, the feature vectors of probe image and the video sequence were further forward to the similarity sub-network for distance metric learning. In the proposed framework, the feature representation and the similarity metric learning can be learned and optimized simultaneously. We evaluated the proposed framework on three public person re-id data sets, and the experimental results showed that the proposed approach is effective for the image-to-video person re-id. [ABSTRACT FROM AUTHOR]

Inter prediction is a crucial part of hybrid video coding frameworks, utilized to exploit the temporal redundancy in video sequences and improve the coding performance. During inter prediction, a predicted block is typically derived from reference pictures using motion estimation and motion compensation. To improve the coding performance of inter prediction, a neural network based enhancement to inter prediction (NNIP) is proposed in this paper. NNIP is composed of three networks, namely residue estimation network, combination network, and deep refinement network. Specifically, first, a residue estimation network is designed to estimate the residue between current block and its predicted block using their available spatial neighbors. Second, the feature maps of the estimated residue and the predicted block are extracted and concatenated in a combination network. Finally, the concatenated feature maps are fed into a deep refinement network to generate a refined residue, which is added back to the predicted block to derive a more accurate predicted block. NNIP is integrated in HEVC to evaluate its efficiency. The experimental results demonstrate that NNIP can achieve 4.6%, 3.0%, and 2.7% BD-rate reduction on average under LDP, LDB, and RA configurations compared to HEVC. [ABSTRACT FROM AUTHOR]

Residual networks (ResNets) have introduced a milestone for the deep learning community due to their outstanding performance in diverse applications. They enable efficient training of increasingly deep networks, reducing the training difficulty and error. The main intuition behind them is that, instead of mapping the input information, they are mapping a residual part of it. Since the original work, a lot of extensions have been proposed to improve information mapping. In this paper, a novel extension of the residual block is proposed inspired by linear dynamical systems (LDSs), called LDS-ResNet. Specifically, a new module is presented that improves mapping of residual information by transforming it in a hidden state and then mapping it back to the desired feature space using convolutional layers. The proposed module is utilized to construct multi-branch residual blocks for convolutional neural networks. An exploration of possible architectural choices is presented and evaluated. Experimental results show that LDS-ResNet outperforms the original ResNet in image classification and object detection tasks on public datasets such as CIFAR-10/100, ImageNet, VOC, and MOT2017. Moreover, its performance boost is complementary to other extensions of the original network such as pre-activation and bottleneck, as well as stochastic training and Squeeze-Excitation. [ABSTRACT FROM AUTHOR]

This paper proposes a deep learning method for low-light image enhancement, which exploits the generation capability of Neural Networks (NNs) while requiring no training samples except the input image itself. Based on the Retinex decomposition model, the reflectance and illumination of a low-light image are parameterized by two untrained NNs. The ambiguity between the two layers is resolved by the discrepancy between the two NNs in terms of architecture and capacity, while the complex noise with spatially-varying characteristics is handled by an illumination-adaptive self-supervised denoising module. The enhancement is done by jointly optimizing the Retinex decomposition and the illumination adjustment. Extensive experiments show that the proposed method not only outperforms existing non-learning-based and unsupervised-learning-based methods, but also competes favorably with some supervised-learning-based methods in extreme low-light conditions. [ABSTRACT FROM AUTHOR]

Face recognition has attracted increasing attention due to its wide range of applications, but it is still challenging when facing large variations in the biometric data characteristics. Lenslet light field cameras have recently come into prominence to capture rich spatio-angular information, thus offering new possibilities for advanced biometric recognition systems. This paper proposes a double-deep spatio-angular learning framework for light field-based face recognition, which is able to model both the intra-view/spatial and inter-view/angular information using two deep networks in sequence. This is a novel recognition framework that has never been proposed in the literature for face recognition or any other visual recognition task. The proposed double-deep learning framework includes a long short-term memory (LSTM) recurrent network, whose inputs are VGG-Face descriptions, computed using a VGG-16 convolutional neural network (CNN). The VGG-Face spatial descriptions are extracted from a selected set of 2D sub-aperture (SA) images rendered from the light field image, corresponding to different observation angles. A sequence of the VGG-Face spatial descriptions is then analyzed by the LSTM network. A comprehensive set of experiments has been conducted using the IST-EURECOM light field face database, addressing varied and challenging recognition tasks. The results show that the proposed framework achieves superior face recognition performance when compared to the state of the art. [ABSTRACT FROM AUTHOR]

In this paper, the problem of human pose estimation in a 2D still image is addressed. A framework called adaptive holons representation (ADORE) that takes advantage of local and global cues is proposed to improve the pose estimation accuracy. In particular, ADORE is made up of two components: 1) the holons part, independent losses pose nets (ILPNs) is designed to first infer joints location on the global level; and 2) the adaptive part, convolutional local detectors (CLDs) is proposed to subsequently detect the joints in the potential regions generated by ILPN. Pose estimation is formulated as a classification problem toward body joints in ILPN, which consists of two independent loss layers that, respectively, instruct the learning of $x$ and $y$ coordinates of a joint. Experimental results on two challenging benchmark tasks demonstrate that our proposed framework is more efficient than other deep models and has desirable performance. [ABSTRACT FROM AUTHOR]

We observed that remarkable and impressive performance on image-based human pose estimation have been achieved by deep Convolutional Neural Networks (CNN). Nevertheless, directly applying these image-based models on videos is not only computionally intensive, but also may cause jitter and loss. The main reason is that the image-based models purely focus on the local features of individual frames and totally ignore the temporal information among adjacent frames. Some existing methods are proposed to address the temporal coherency issue. However, these methods need to be designed carefully and cannot be combined with existing image-based methods. In this paper, we propose a simple yet effective module to refine the estimated pose by exploiting the temporal coherency among the heatmaps of adjacent frames, which can be easily inserted into image-based networks as a plug-in. We show that the temporal coherency issue among the heatmap frames could be re-formulated as a graph path selection optimization problem. Moreover, to speed up the refinement process, we propose a hierarchical graph optimization to achieve the refinement from coarse to fine. Experimental results on two large-scale video pose estimation benchmarks show that our module can improve the performance with little speed loss when combined with image-based methods as an efficient plug-in. [ABSTRACT FROM AUTHOR]

Due to occlusion among views and temporal inconsistency in depth video, spatio-temporal distortion occurs in 3D synthesized video with depth image-based rendering. In this paper, we propose a deep Convolutional Neural Network (CNN)-based synthesized video denoising algorithm to reduce temporal flicker distortion and improve perceptual quality of 3D synthesized video. First, we analyze the spatio-temporal distortion, and model eliminating spatio-temporal distortion as a perceptual video denoising problem. Then, a deep learning-based synthesized video denoising network is proposed, in which a CNN-friendly spatio-temporal loss function is derived from a synthesized video quality metric and integrated with a single image denoising network architecture. Finally, specific schemes, i.e., specific Synthesized Video Denoising Networks (SynVD-Nets), and a general scheme, i.e., General SynVD-Net (GSynVD-Net), based on existing CNN-based denoising models, are developed to handle synthesized video with different distortion levels more effectively. Experimental results show that the proposed SynVD-Net and GSynVD-Net can outperform deep learning-based counterparts and conventional denoising methods, and significantly enhance perceptual quality of 3D synthesized video. [ABSTRACT FROM AUTHOR]

Convolutional neural networks (CNNs) demonstrate excellent performance in various computer vision applications. In recent years, FPGA-based CNN accelerators have been proposed for optimizing performance and power efficiency. Most accelerators are designed for object detection and recognition algorithms that are performed on low-resolution images. However, real-time image super-resolution (SR) cannot be implemented on a typical accelerator because of the long execution cycles required to generate high-resolution (HR) images, such as those used in ultra-high-definition systems. In this paper, we propose a novel CNN accelerator with efficient parallelization methods for SR applications. First, we propose a new methodology for optimizing the deconvolutional neural networks (DCNNs) used for increasing feature maps. Second, we propose a novel method to optimize CNN dataflow so that the SR algorithm can be driven at low power in display applications. Finally, we quantize and compress a DCNN-based SR algorithm into an optimal model for efficient inference using on-chip memory. We present an energy-efficient architecture for SR and validate our architecture on a mobile panel with quad-high-definition resolution. Our experimental results show that, with the same hardware resources, the proposed DCNN accelerator achieves a throughput up to 108 times greater than that of a conventional DCNN accelerator. In addition, our SR system achieves an energy efficiency of 144.9, 293.0, and 500.2 GOPS/W at SR scale factors of 2, 3, and 4, respectively. Furthermore, we demonstrate that our system can restore HR images to a high quality while greatly reducing the data bit-width and the number of parameters compared with conventional SR algorithms. [ABSTRACT FROM AUTHOR]

In general, an image attribute is a human-nameable visual property that has a semantic connotation. Appropriate modeling of the intrinsic contextual correlations among attributes plays a fundamental role in attribute detection. In this paper, we consider image attribute detection from the perspective of regularized deep learning. In particular, we propose a regularized deep belief network (rDBN) to perform the image attribute detection task, which is composed of two parts: 1) a detection DBN (dDBN) that models the joint distribution of images and their corresponding attributes, which acts as an attribute detector and 2) a contextual restricted Boltzmann machine that explicitly models the correlations among attributes acting as a regularizer that restraints the output detection result given by the dDBN to meet the contextual prior of attributes. Furthermore, we propose an efficient fine-tuning scheme that can further optimize the performance of the dDBN by backpropagation. Experimental results show that the proposed rDBN obtains improvements over the state-of-the-art methods for attribute detection on the benchmark data sets. [ABSTRACT FROM PUBLISHER]

Recently, deep learning techniques have substantially boosted the performance of salient object detection in still images. However, the salient object detection in videos by using traditional handcrafted features or deep learning features is not fully investigated, probably due to the lack of sufficient manually labeled video data for saliency modeling, especially for the data-driven deep learning. This paper proposes a novel weakly supervised approach to the salient object detection in a video, which can learn a robust saliency prediction model by using very limited manually labeled data and a large amount of weakly labeled data that could be easily generated in a supervised approach. Furthermore, we propose a spatiotemporal cascade neural network architecture for saliency modeling, in which two fully convolutional networks are cascaded to evaluate the visual saliency from both spatial and temporal cues to lead the optimal video saliency prediction. The proposed approach is extensively evaluated on the widely used challenging data sets, and the experiments demonstrate that our proposed approach substantially outperforms the state-of-the-art salient object detection models. [ABSTRACT FROM AUTHOR]

A number of visual tracking methods achieve the state-of-the-art performance based on deep learning recently. However, most of these trackers utilize the deep neural network in regression task or classification task separately. In this paper, we propose an adversarial deep tracking framework. The framework is composed of a fully convolutional Siamese neural network (regression network) and a discriminative classification network. Then, we jointly optimize the regression network and the classification network by adversarial learning. In the uniform framework, the regression network and classification network can be trained end-to-end as a whole using large amounts of video training data sets. During the testing phase, the regression network generates a response map which reflects the location and the size of the target within each candidate search patch, and the classification network discriminates which response map is the best in terms of the corresponding template patch and candidate search patch. In addition, we propose an attention visualization algorithm for our tracker, and it reflects the area that attracts the attention of our tracker during tracking. The experimental results on three large-scale visual tracking benchmarks (OTB-100, TC-128, and VOT2016) demonstrate the effectiveness of the proposed tracking algorithm and show that our tracker performs comparably against the state-of-the-art trackers. [ABSTRACT FROM AUTHOR]

In this paper, we propose a deep convolutional neural network (DCNN) architecture that addresses the problem of video/image concept annotation by exploiting concept relations at two different levels. At the first level, we build on ideas from multi-task learning, and propose an approach to learn concept-specific representations that are sparse, linear combinations of representations of latent concepts. By enforcing the sharing of the latent concept representations, we exploit the implicit relations between the target concepts. At a second level, we build on ideas from structured output learning and propose the introduction, at training time, of a new cost term that explicitly models the correlations between the concepts. By doing so, we explicitly model the structure in the output space (i.e., the concept labels). Both of the above are implemented using standard convolutional layers and are incorporated in a single DCNN architecture that can then be trained end-to-end with standard back-propagation. Experiments on four large-scale video and image data sets show that the proposed DCNN improves concept annotation accuracy and outperforms the related state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Depth estimation from single monocular images is a key component in scene understanding. Most existing algorithms formulate depth estimation as a regression problem due to the continuous property of depths. However, the depth value of input data can hardly be regressed exactly to the ground-truth value. In this paper, we propose to formulate depth estimation as a pixelwise classification task. Specifically, we first discretize the continuous ground-truth depths into several bins and label the bins according to their depth ranges. Then, we solve the depth estimation problem as classification by training a fully convolutional deep residual network. Compared with estimating the exact depth of a single point, it is easier to estimate its depth range. More importantly, by performing depth classification instead of regression, we can easily obtain the confidence of a depth prediction in the form of probability distribution. With this confidence, we can apply an information gain loss to make use of the predictions that are close to ground-truth during training, as well as fully-connected conditional random fields for post-processing to further improve the performance. We test our proposed method on both indoor and outdoor benchmark RGB-Depth datasets and achieve state-of-the-art performance. [ABSTRACT FROM AUTHOR]

The state-of-the-art performance for object detection has been significantly improved over the past two years. Besides the introduction of powerful deep neural networks, such as GoogleNet and VGG, novel object detection frameworks, such as R-CNN and its successors, Fast R-CNN, and Faster R-CNN, play an essential role in improving the state of the art. Despite their effectiveness on still images, those frameworks are not specifically designed for object detection from videos. Temporal and contextual information of videos are not fully investigated and utilized. In this paper, we propose a deep learning framework that incorporates temporal and contextual information from tubelets obtained in videos, which dramatically improves the baseline performance of existing still-image detection frameworks when they are applied to videos. It is called T-CNN, i.e., tubelets with convolutional neueral networks. The proposed framework won newly introduced an object-detection-from-video task with provided data in the ImageNet Large-Scale Visual Recognition Challenge 2015. Code is publicly available at https://github.com/myfavouritekk/T-CNN. [ABSTRACT FROM AUTHOR]

Deep learning, e.g., convolutional neural networks (CNNs), has achieved great success in image processing and computer vision especially in high-level vision applications, such as recognition and understanding. However, it is rarely used to solve low-level vision problems such as image compression studied in this paper. Here, we move forward a step and propose a novel compression framework based on CNNs. To achieve high-quality image compression at low bit rates, two CNNs are seamlessly integrated into an end-to-end compression framework. The first CNN, named compact convolutional neural network (ComCNN), learns an optimal compact representation from an input image, which preserves the structural information and is then encoded using an image codec (e.g., JPEG, JPEG2000, or BPG). The second CNN, named reconstruction convolutional neural network (RecCNN), is used to reconstruct the decoded image with high quality in the decoding end. To make two CNNs effectively collaborate, we develop a unified end-to-end learning algorithm to simultaneously learn ComCNN and RecCNN, which facilitates the accurate reconstruction of the decoded image using RecCNN. Such a design also makes the proposed compression framework compatible with existing image coding standards. Experimental results validate that the proposed compression framework greatly outperforms several compression frameworks that use existing image coding standards with the state-of-the-art deblocking or denoising post-processing methods. [ABSTRACT FROM AUTHOR]

Emotion recognition is challenging due to the emotional gap between emotions and audio–visual features. Motivated by the powerful feature learning ability of deep neural networks, this paper proposes to bridge the emotional gap by using a hybrid deep model, which first produces audio–visual segment features with Convolutional Neural Networks (CNNs) and 3D-CNN, then fuses audio–visual segment features in a Deep Belief Networks (DBNs). The proposed method is trained in two stages. First, CNN and 3D-CNN models pre-trained on corresponding large-scale image and video classification tasks are fine-tuned on emotion recognition tasks to learn audio and visual segment features, respectively. Second, the outputs of CNN and 3D-CNN models are combined into a fusion network built with a DBN model. The fusion network is trained to jointly learn a discriminative audio–visual segment feature representation. After average-pooling segment features learned by DBN to form a fixed-length global video feature, a linear Support Vector Machine is used for video emotion classification. Experimental results on three public audio–visual emotional databases, including the acted RML database, the acted eNTERFACE05 database, and the spontaneous BAUM-1s database, demonstrate the promising performance of the proposed method. To the best of our knowledge, this is an early work fusing audio and visual cues with CNN, 3D-CNN, and DBN for audio–visual emotion recognition. [ABSTRACT FROM AUTHOR]

Recently considerable efforts have been dedicated to unconstrained face recognition, which requires to identify faces “in the wild” for a set of images and/or video frames captured without human intervention. Unlike traditional face recognition that compares one-to-one media (either a single image or a video frame) only, we encounter a problem of matching sets with heterogeneous contents containing both images and videos. In this paper, we propose a novel set-to-set (S2S) distance measure to calculate the similarity between two sets with the aim to improve the recognition accuracy for faces with real-world challenges, such as extreme poses or severe illumination conditions. Our S2S distance adopts the kNN-average pooling for the similarity scores computed on all the media in two sets, making the identification far less susceptible to the poor representations (outliers) than traditional feature-average pooling and score-average pooling. Furthermore, we show that various metrics can be embedded into our S2S distance framework, including both predefined and learned ones. This allows to choose the appropriate metric depending on the recognition task in order to achieve the best results. To evaluate the proposed S2S distance, we conduct extensive experiments on the challenging set-based IJB-A face data set, which demonstrate that our algorithm achieves the state-of-the-art results and is clearly superior to the baselines, including several deep learning-based face recognition algorithms. [ABSTRACT FROM AUTHOR]

In this paper, we tackle for the first time, the problem of self-supervised representation learning for free-hand sketches. This importantly addresses a common problem faced by the sketch community – that annotated supervisory data are difficult to obtain. This problem is very challenging in which sketches are highly abstract and subject to different drawing styles, making existing solutions tailored for photos unsuitable. Key for the success of our self-supervised learning paradigm lies with our sketch-specific designs: (i) we propose a set of pretext tasks specifically designed for sketches that mimic different drawing styles, and (ii) we further exploit the use of the textual convolution network (TCN) together with the convolutional neural network (CNN) in a dual-branch architecture for sketch feature learning, as means to accommodate the sequential stroke nature of sketches. We demonstrate the superiority of our sketch-specific designs through two sketch-related applications (retrieval and recognition) on a million-scale sketch dataset, and show that the proposed approach outperforms the state-of-the-art unsupervised representation learning methods, and significantly narrows the performance gap between with supervised representation learning. 1 PyTorch code of this work is available at https://github.com/zzz1515151/self-supervised_learning_sketch. [ABSTRACT FROM AUTHOR]

In this paper, we propose a deep metric learning (DML) approach for robust visual tracking under the particle filter framework. Unlike most existing appearance-based visual trackers, which use hand-crafted similarity metrics, our DML tracker learns a nonlinear distance metric to classify the target object and background regions using a feed-forward neural network architecture. Since there are usually large variations in visual objects caused by varying deformations, illuminations, occlusions, motions, rotations, scales, and cluttered backgrounds, conventional linear similarity metrics cannot work well in such scenarios. To address this, our proposed DML tracker first learns a set of hierarchical nonlinear transformations in the feed-forward neural network to project both the template and particles into the same feature space where the intra-class variations of positive training pairs are minimized and the interclass variations of negative training pairs are maximized simultaneously. Then, the candidate that is most similar to the template in the learned deep network is identified as the true target. Experiments on the benchmark data set including 51 challenging videos show that our DML tracker achieves a very competitive performance with the state-of-the-art trackers. [ABSTRACT FROM AUTHOR]

Learning-based face descriptors have constantly improved the face recognition performance. Compared with the hand-crafted features, learning-based features are considered to be able to exploit information with better discriminative ability for specific tasks. Motivated by the recent success of deep learning, in this paper, we extend the original shallow face descriptors to deep discriminant face features by introducing a stacked image descriptor (SID). With deep structure, more complex facial information can be extracted and the discriminant and compactness of feature representation can be improved. The SID is learned in a forward optimization way, which is computational efficient compared with deep learning. Extensive experiments on various face databases are conducted to show that SID is able to achieve high face recognition performance with compact face representation, compared with other state-of-the-art descriptors. [ABSTRACT FROM AUTHOR]

The informativeness and discriminativeness of features collaboratively ensure high-accuracy Facial Expression Recognition (FER) in the wild. Most of existing methods use the single-path deep convolutional neural network with softmax loss for basic FER, while they cannot deal with the challenging situations of the compound FER in the wild, because they fail to learn informative and discriminative features in a targeted manner. To this end, we present an Informative and Discriminative Feature Learning (IDFL) framework that consists of two key components: the Multi-Path Attention Convolutional Neural Network (MPACNN) and Balanced Separate loss (BS loss), for both basic and compound high-accuracy FER in the wild. Specifically, MPACNN leverages different paths to learn diverse features. These features are then adaptively fused into informative ones via an attention module, such that the model can adequately capture detailed information for both basic and compound FER. The BS loss maximizes the inter-class distance of features and minimizes the intra-class one. In this way, the features are discriminative enough for high-accuracy FER in the wild. Particularly, the BS loss is invoked as the objective function of MPACNN, so the model can learn informative and discriminative features at the same time, yielding better performance. Seven databases are utilized to evaluate the proposed method, and the results demonstrate that our method achieves state-of-the-art performance on both basic and compound expressions with good generalization ability. Moreover, our model contains fewer parameters and can be trained faster than other related models. [ABSTRACT FROM AUTHOR]

Person reidentification involves recognizing a person across nonoverlapping camera views, with different pose, illumination, and camera characteristics. We propose to tackle this problem by training a deep convolutional network to represent a person’s appearance as a low-dimensional feature vector that is invariant to common appearance variations encountered in the reidentification problem. Specifically, a Siamese network architecture is used to train a feature extraction network using pairs of similar and dissimilar images. We show that the use of a novel multitask learning objective is crucial for regularizing the network parameters in order to prevent overfitting due to the small size of the training data set. We complement the verification task, which is at the heart of reidentification, by training the network to jointly perform verification and identification and to recognize attributes related to the clothing and pose of the person in each image. In addition, we show that our proposed approach performs well even in the challenging cross-data set scenario, which may better reflect real-world expected performance. [ABSTRACT FROM PUBLISHER]

The deep convolutional neural network (CNN) based tracking-by-detection framework recently has become one of the most popular trackers. However, these methods are either time consuming or have greatly reduced performance. This article aims to achieve nearly identical tracking accuracy with the state-of-the-art CNN tracking-by-detection algorithm with relatively faster speed. We study the existing excellent trackers under the CNN tracking-by-detection framework and introduce the following: a multiscale feature pyramid fusion neural network based on dilated convolutions is constructed to learn a scale-invariant discriminative representation for tracking small objects, a hard-threshold weighted cross-entropy loss function is proposed to decrease the gap between object classification and tracking, and a mutual learning-based training policy is used to fuse the information from the network trained by image patches with different contextual regions to further improve the tracking performance. We conduct comprehensive experiments on visual object tracking benchmarks that validate the achievement of competitive performance of the proposed tracker with relatively faster speed both on qualitative and quantitative criteria. Additionally, the experimental results reveal that the proposed mutual learning-based training policy can accelerate the convergence speed and achieve better generalization performance. [ABSTRACT FROM AUTHOR]

The human vision system usually has a specifically activated area of neurons when recognizing a category of images. Inspired by this visual mechanism, we propose a layerwise class-aware convolutional neural network architecture to explicitly discover category-tailored neurons on intermediate hidden layers to improve the network learning ability. Instead of directly selecting activated neurons for different categories, we inversely suppress those neurons of intermediate layers irrelevant with the given target class to produce a class-specific subnetwork, which implicitly enhances the discriminability of hidden layer features due to the increase of the inter-class discrepancy on them. Together with the classifier of the top layer, we jointly learn this network by formulating the suppressor of hidden layers as a penalty term in the objective function. To address class-specific neuron suppression in each hidden layer, we also introduce a statistic method based on mutual information to dynamically and automatically update the suppressed neurons during the network training. Extensive experiments demonstrate that the proposed model is superior to the state-of-the-art models. [ABSTRACT FROM AUTHOR]

We present a novel, deep Q-learning based method, QLP, for pruning deep neural networks (DNNs). Given a DNN, our method intelligently determines favorable layer-wise sparsity ratios, which are then implemented via unstructured, magnitude-based, weight pruning. In contrast to previous reinforcement learning (RL) based pruning methods, our method is not forced to prune a DNN within a single, sequential pass from the first layer to the last. It visits each layer multiple times and prunes them little by little at each visit, achieving superior granular pruning. Moreover, our method is not restricted to a subset of actions within the feasible action space. It has the flexibility to execute a whole range of sparsity ratios (0% - 100%) for each layer. This enables aggressive pruning without compromising accuracy. Furthermore, our method does not require a complex state definition; it features a simple, generic definition that is composed of only the index and the density of the layers, which leads to less computational demand while observing the state at each interaction. Lastly, our method utilizes a carefully designed curriculum that enables learning targeted policies for each sparsity regime, which helps to deliver better accuracy, especially at high sparsity levels. We conduct batched performance tests at compelling sparsity levels (up to 98%), present extensive ablation studies to justify our RL-related design choices, and compare our method with the state-of-the-art, including RL-based and other pruning methods. Our method sets the new state-of-the-art results in most of the experiments with ResNet-32 and ResNet-56 over CIFAR-10 dataset as well as ResNet-50 and MobileNet-v1 over ILSVRC2012 (ImageNet) dataset. [ABSTRACT FROM AUTHOR]

In this work, we propose to push the performance limit of semi-supervised learning on very small sets of labeled samples by developing a new method called learned model composition with critical sample look-ahead (LMCS). Training efficient deep neural networks on much smaller sets of labeled samples is a challenging problem. With a small labeled set, the initial network suffers from low accuracy. Based on this error-prone network, the subsequent semi-supervised learning process will be fragile and unstable. To address this issue, we propose to introduce a look-ahead master model to identify the correct direction of model evolution to effectively guide the semi-supervised learning process of the student model. Specifically, our proposed LMCS method explores two major ideas. First, it introduces a new learned model composition structure so that we can compose a more efficient master network from student models of past iterations through a network learning process. Second, we develop a new method, called confined maximum entropy search, to discover new critical samples near the model decision boundary and provide the master model with look-ahead access to these samples to enhance its guidance capability. Our extensive experimental results demonstrate that the proposed LMCS method outperforms the state-of-the-art semi-supervised learning methods, especially on small sets of labeled samples. For example, on the CIFAR-10 dataset, with a very small set of 80 labeled samples, our method outperforms Google’s MixMatch method, reducing the error rate by more than 10%. [ABSTRACT FROM AUTHOR]

Two approaches are proposed for cross-pose face recognition, one is built on the handcrafted features extracted from the 3D reconstruction of facial components and the other is built on the learned features from a deep convolutional neural network (CNN). As both approaches rely on facial landmarks for alignment across large poses, we propose the Fast Hierarchical Model (FHM) for locating cross-pose facial landmarks in real time. Unlike most 3D approaches that consider holistic faces, the first proposed approach considers 3D facial components. It segments each 2D face in the gallery into components, reconstructs the 3D surface for each component, and recognizes a query face by component features. The core part of the CNN-based approach is a modified VGG network. We study the performance with different settings on the training set, including the synthesized data from 3D reconstruction, the real-life data from an in-the-wild database, and both types of data combined. The two recognition approaches and the FHM are evaluated in extensive experiments and compared with state-of-the-art methods to demonstrate their efficacy. [ABSTRACT FROM AUTHOR]

This letter presents an effective method to encode the spatiotemporal information of a skeleton sequence into color texture images, referred to as skeleton optical spectra, and employs convolutional neural networks (ConvNets) to learn the discriminative features for action recognition. Such spectrum representation makes it possible to use a standard ConvNet architecture to learn suitable “dynamic” features from skeleton sequences without training millions of parameters afresh and it is especially valuable when there is insufficient annotated training video data. Specifically, the encoding consists of four steps: mapping of joint distribution, spectrum coding of joint trajectories, spectrum coding of body parts, and joint velocity weighted saturation and brightness. Experimental results on three widely used datasets have demonstrated the efficacy of the proposed method. [ABSTRACT FROM PUBLISHER]