Search

Device activity detection has been extensively investigated for grant-free massive machine-type communications. Instead of using deep Multi-Layer Perception (MLP) networks, this letter proposes a novel convolutional neural network (CNN) architecture for learning device activity from multiple-measurement vectors (MMV). With the use of $1\times 1$ convolutional layers, the proposed CNN could exploit the full potential of shared sparsity among multiple measurements. Extensive simulations show that the proposed CNN outperforms the existing deep MLP network in both performance and computational complexity, especially when the number of measurements increases. more...

Visual saliency detection plays a significant role in the fields of computer vision. In this paper, we introduce a novel saliency detection method based on weighted linear multiple kernel learning (WLMKL) framework, which is able to adaptively combine different contrast measurements in a supervised manner. As most influential factor is contrast operation in bottom-up visual saliency, an average weighted corner-surround contrast (AWCSC) is first designed to measure local visual saliency. Combined with common-used center-surrounding contrast (CESC) and global contrast (GC), three types of contrast operations are fed into our WLMKL framework to produce the final saliency map. We show that the assigned weights for each contrast feature maps are always normalized in our WLMKL formulation. In addition, the proposed approach benefits from the advantages of the contribution of each individual contrast feature maps, yielding more robust and accurate saliency maps. We evaluated our method for two main visual saliency detection tasks: human fixed eye prediction and salient object detection. The extensive experimental results show the effectiveness of the proposed model, and demonstrate the integration is superior than individual subcomponent. more...

Recently, exploring features from different layers in fully convolutional networks (FCNs) has gained substantial attention to capture context information for semantic segmentation. This paper presents a novel encoder-decoder architecture, called contextual ensemble network (CENet), for semantic segmentation, where the contextual cues are aggregated via densely usampling the convolutional features of deep layer to the shallow deconvolutional layers. The proposed CENet is trained in terms of end-to-end segmentation to match the resolution of input image, and allows us to fully explore contextual features through ensemble of dense deconvolutions. We evaluate our CENet on two widely-used semantic segmentation datasets: PASCAL VOC 2012 and CityScapes. The experimental results demonstrate our CENet achieves superior performance with respect to recent state-of-the-art results. Furthermore, we also evaluate CENet on MS COCO dataset and ISBI 2012 dataset for the task of instance segmentation and biological segmentation, respectively. The experimental results show that CENet obtains promising results on these two datasets. more...

Open set domain adaptation by back propagation (OSDA-BP) was recently proposed as a novel end-to-end training approach for tackling the open set scenario, where only a few categories of interest are shared between source and target data. This paper provides an insightful understanding of the binary cross entropy loss employed in OSDA-BP for picking up the potential unknown samples. With this new understanding, we propose to replace the binary cross entropy loss with a symmetrical Kullback Leibler(KL) distance based loss. This improved OSDA-BP method is extensively evaluated over Office-31 dataset and a considerable performance improvement is observed. more...

This paper presents a novel saliency detection method based on weighted linear multiple kernel learning (WLMKL), which is able to adaptively combine different contrast measurements in a supervised manner. Three commonly used bottom-up visual saliency operations are first introduced, including corner-surround contrast (CSC), center-surround contrast (CESC), and global contrast (GC). Then these contrast measures are fed into our WLMKL framework to produce the final saliency map. We show that the assigned weights for each contrast feature maps are always normalized in our WLMKL formulation. In addition, the proposed approach benefits from the advantages of the contribution of each individual contrast operation, and thus produces more robust and accurate saliency maps. The extensive experimental results show the effectiveness of the proposed model, and demonstrate the combination is superior to individual subcomponent. more...

Fully convolutional neural networks (FCNs) have shown great success in dense estimation tasks. One key pillar of such progress is mining multi-scale context cues from features in different convolutional layers. This paper introduces contextual aggregating network(CAN), a generic convolutional feature ensembling framework for semantic segmentation. Our framework first captures multi-scale contextual clues by concatenating multi-level feature representation, which carries both coarse semantics and fine details. Then it adaptively integrates stacked features to perform dense pixel estimation. The proposed CAN is trainable end-to-end, and allows us to fully investigate multi-scale context information embedded in images. The experiments show the promising results of our method on PASCAL VOC 2012 and Cityscapes dataset. more...

The grayscale-thermal tracking has attracted increasing attention due to the fact that it can make thermal information complement with grayscale information. Since there exists a large gap between the grayscale and the thermal video sequences, how to exploit the intrinsic relation between the grayscale and the thermal targets has become the key point. To address this issue, in this paper, we propose an inverse sparse representation based framework for the grayscale-thermal tracking, in which a canonical correlation analysis based inverse sparse representation model is adopted to jointly encode the target candidates in the grayscale and the thermal video sequences. The target coding process can explore the similarity between the grayscale and the thermal appearance in a common subspace, which can highlight the useful and discriminative information in both grayscale and thermal targets. The experiments on OSU-CT dataset can illustrate the promising performance of our tracking framework. more...

Recently, exploring multiple feature maps from different layers in fully convolutional networks (FCNs) has gained substantial attention to capture context information for semantic segmentation. This paper presents a novel encoder-decoder architecture, called dense deconvolutional network (DDN), for semantic segmentation, where the feature maps of deeper convolutional layers are densely upsampled for the shallow deconvolutional layers. The proposed DDN is trainable end-to-end, and allows us to fully investigate multiple scale context cues embedded in images. The experimental results show that our DDN outperforms previous FCNs and encoder-decoder networks (EDNs) on PASCAL VOC 2012 dataset. more...

The state-of-the-art training of deep neural networks requires to normalize the activities of the neurons for accelerating the training process. A standard approach is to employ batch normalization (BN), in which the activations are normalized by the mean and standard deviation of the training mini-batch. To be invertible, BN also introduces an adaptive gain and bias which are applied after the normalization but often before the non-linearity. In this paper, we investigate the effects of learnable parameters, gain and bias, on the training of various typical deep neural nets, including ALL-CNNs, Network In Network (NIN), ResNets. Through extensive experiments, we show that there is no big difference in both training convergence and final test accuracy if we remove the BN layer following the final convolutional layer from a convolutional neural network (CNN) for standard classification tasks. We also observed that without adaptively updating learnable parameters for BN layers, it often requires less time for training of very deep neural nets such as ResNet-101. more...

In traditional sparse representation based visual tracking, the particles are densely sampled, the appearance of some candidates may be very similar, hence the particle observations can be divided into disjointed groups. Existing methods only exploit the group similarity in a certain feature space. In this paper we propose a multi-view discriminant based multitask sparse representation method to exploit the group similarity in a multi-feature space. The proposed method can discriminate the reliability of observation groups and achieve a proper multi-view fusion by using a multi-view discriminant matrix to project multi-feature observation groups into a common subspace. Experiment results show that our method can achieve a better tracking performance than state-of-the-art tracking methods do. more...

This paper explores the performance of Neural Network (NN) for human action recognition and proposes a novel hierarchical and boosting-based action recognition system. Specifically, the main contributions of our work are three-fold: (1) A boosted NN based scheme is applied to the human action recognition task for the first time, during which we extend the standard binary AdaBoost algorithm to a multiclass version; (2) A novel hierarchical recognition framework with pre-decision and post-decision modules is proposed, which can significantly enhance the training efficiency as well as the frame-based recognition accuracy; (3) Numerous modified features (both motion and shape features) are utilized and combined in this paper. Experiments on the Weizmann dataset show promising results of our approach in comparison with other state-of-the-art methods. more...

In this paper, we address the problem of using boosting to detect different classes of objects with significant intra-class variation. Current approaches generally require prior knowledge such as semantic information to obtain an explicit partition over positive samples. This can require a lot of manually labeled samples and can limit the performance of classifier due to subjective error. We present a novel JointBoost based learning method which can learn the variant appearances of targets without any prior assumption. With an implicit partition, the specific features for fractional samples can contribute to classification as well as generic features for the whole training set. By encouraging an intra-class and inter-class feature sharing between implicit sub-categories, our data-driven learning approach avoids a local optimum in the candidate weak classifier space. Experimental results on two popular tasks demonstrate the considerable improvements brought by the new approach. We hope that implicit JointBoost will extend the application of traditional boosting methods. more...