Search

3D reconstruction from single-view images is a long-standing research problem. To date, there are various methods based on point clouds and volumetric representations. In spite of success in 3D models generation, it is quite challenge for these approaches to deal with models with complex topology and fine geometric details. Thanks to the recent advance of deep shape representations,learning the structure and detail representation using deep neural networks is a promising direction. In this paper, we propose a novel approach named STD-Net to reconstruct 3D models utilizing mesh representation that is well suited for characterizing complex structures and geometry details. Our method consists of (1) an auto-encoder network for recovering the structure of an object with bounding box representation from a single-view image; (2) a topology-adaptive GCN for updating vertex position for meshes of complex topology; and (3) a unified mesh deformation block that deforms the structural boxes into structure-aware meshes. Evaluation on ShapeNet shows that STD-Net has better performance than state-of-the-art methods in reconstructing complex structures and fine geometric details.

Studying the deep learning-based molecular representation has great significance on predicting molecular property, promoted the development of drug screening and new drug discovery, and improving human well-being for avoiding illnesses. It is essential to learn the characterization of drug for various downstream tasks, such as molecular property prediction. In particular, the 3D structure features of molecules play an important role in biochemical function and activity prediction. The 3D characteristics of molecules largely determine the properties of the drug and the binding characteristics of the target. However, most current methods merely rely on 1D or 2D properties while ignoring the 3D topological structure, thereby degrading the performance of molecular inferring. In this paper, we propose 3DMol-Net to enhance the molecular representation, considering both the topology and rotation invariance (RI) of the 3D molecular structure. Specifically, we construct a molecular graph with soft relations related to the spatial arrangement of the 3D coordinates to learn 3D topology of arbitrary graph structure and employ an adaptive graph convolutional network to predict molecular properties and biochemical activities. Comparing with current graph-based methods, 3DMol-Net demonstrates superior performance in terms of both regression and classification tasks. Further verification of RI and visualization also show better robustness and representation capacity of our model.

Small-object detection has long been a challenge. High-megapixel cameras are used to solve this problem in industries. However, current detectors are inefficient for high-resolution images. In this work, we propose a new module called Pre-Locate Net, which is a plug-and-play structure that can be combined with most popular detectors. We inspire the use of classification ideas to obtain candidate regions in images, greatly reducing the amount of calculation, and thus achieving rapid detection in high-resolution images. Pre-Locate Net mainly includes two parts, candidate region classification and behavior classification. Candidate region classification is used to obtain a candidate region, and behavior classification is used to estimate the scale of an object. Different follow-up processing is adopted according to different scales to balance the variance of the network input. Different from the popular candidate region generation method, we abandon the idea of regression of a bounding box and adopt the concept of classification, so as to realize the prediction of a candidate region in the shallow network. We build a high-resolution dataset of aircraft and landing gears covering complex scenes to verify the effectiveness of our method. Compared to state-of-the-art detectors (e.g., Guided Anchoring, Libra-RCNN, and FASF), our method achieves the best mAP of 94.5 on 1920 × 1080 images at 16.7 FPS.

Over the past decades, video salient object detection has achieved rapid development. However, the effect of edge information on spatial and temporal features extraction has not been explored. The method STEG-Net is proposed to solve the problem, we use the extracted edge information to guide the extrction of spatial and temporal features simultaneously. We combine deep texture information with shallow edge information, which can not only retain the edge of the object but also enhance the global information, leading to the more accurate location of the object. At the same time, the shallow edge information and the deep texture information are complementary. As a result, edge information and salient features could advance each other while extracting features. Finally, we evaluate the 16 state-of-the-art VSOD methods on ViSal and FBMS datasets. At the same time we also evaluate the 13 state-of-the-art VSOD methods on UVSD and MCL datasets. The experiment results indicate that the proposed method outperforms the state-of-the-art approaches.

Nowadays, a large number of sensors are equipped on mobile or stationary platforms, which continuously generate geo-tagged and time-stamped readings (i.e., geo-sensory data) that contain rich information about the surrounding environment. These data have irregular space and time coordinates. To represent geo-sensory data, there have been extensive research efforts using time sequences, grid-like images, and graph signals. However, there still lacks a proper representation that can describe both the mobile and stationary geo-sensory data without the information-losing discretization in spatial and temporal dimensions. In this paper, we propose to represent massive geo-sensory data as spatio-temporal point clouds (STPC), and present STPC-Net, a novel deep neural network for processing STPC. STPC leverages the original irregular space-time coordinates, and STPC-Net captures intra-sensor and inter-sensor correlations from STPC. In this way, STPC-Net learns the key information of STPC, and overcomes challenges in data irregularity. Experiments using real-world datasets show that STPC-Net achieves state-of-the-art performance in different tasks on both mobile and stationary geo-sensory data. The source code is available at https://github.com/zhengchuanpan/STPC-Net.

With the rapid development of sensor technologies, time series data collected by multiple and spatially distributed sensors have been widely used in different research fields. Examples of such data include geo-tagged temperature data collected by temperature sensors, air pollutant monitoring data, and traffic data collected by road traffic sensors. Due to sensor failure, communication errors and storage loss, etc., data collected by sensors inevitably includes missing data. However, models commonly used in the analysis of such large-scale data often rely on complete data sets. This paper proposes a model for the imputation of missing data of traffic flow, which combines a self-attention mechanism, an auto-encoder, and a generative adversarial network, into a self-attention generative adversarial imputation net (SA-GAIN). The introduction of the self-attention mechanism can help the proposed model to effectively capture correlations between spatially-distributed sensors at different time points. Adversarial training through two neural networks, called generators and discriminators, allows the proposed model to generate imputed data close to the real data. In comparison with different imputation models, the proposed model shows the best performance in imputing missing data.

Energy-efficient resource-optimization for multiple application scenarios requires appropriate clustering-based network-fractioning that skillfully removes unbalanced network connectivity, which is a stimulating bottleneck in WSN-performance. This paper proposes a new Uniform Connectivity-based clustering Protocol, called Lines-of-Uniformity based Enhanced-Threshold (LUET) to provide energy-efficient coverage in three-tier heterogeneous WSN. This protocol considers the node’s remnant-energy and its proximity from either of the lines of uniformity for lowering down average isolated node count in WSN. This paper also proposes a rotation epoch-based LUET variant, (LUET|R) that incorporates static epoch for initial clustering rounds until First-Node-Death to overcome the rapid fall after the death of the first node. The simulation model demonstrates the superiority of LUET and its variants over other established clustering protocols in terms of network-lifetime, power-efficiency, net death-rate, average isolated-nodes, throughput.

Face aging has attracted widespread attention in recent years, but most studies are based on the same emotional situation. Is the same person’s aging in different emotional situations the same? To solve the above confusion, this paper proposes a novel face aging model DEF-Net, which consists of two parts: different emotional learnings (Emotion-Net) and face aging (Age-Net). Given a target emotion category, DEF-Net first assists the image from the original dataset to learn the emotion features through Emotion-Net and the generated dataset is used as the inputs of Age-Net. At the same time, multiple loss functions are used to ensure that the crucial information of the original image is not lost. Secondly, Age-Net, which has been pre-trained on the original dataset, began to adopt the generated dataset to learn the aging distribution under different emotions. Designed loss functions are utilized to ensure that the realistic target images generated by Age-Net do not lose the learned emotional characteristics. Finally, extensive experiments are used to verify the performance of DEF-Net. Compared with other state-of-the-art methods: (1) DEF-Net can learn different facial emotions across different datasets and generate corresponding realistic aging images; (2) the results achieved by our DEF-Net are demonstrated to be better than those by the model that performs face aging first and then learns different emotional characteristics.

Accurate and in-time performance monitoring plays a great role in industrial processes. However, since the labeled performances are usually measured by some special devices with a relatively long-time interval, the current deep neural networks (DNNs) for the performance monitoring always treat them only as an output. But actually, in the industrial process like froth flotation, labeled performance at the previous moment could also offer valuable information for performance monitoring at the current moment, especially under unstable working conditions. Therefore, we propose a Siamese time series and difference network (STS-D net) which integrates input features at different time steps and labeled performance at the previous moment effectively. The proposed STS-D net includes two sub-networks. One is the Siamese time series network (STS-net) which aims to extract effective and uniform feature representations for the input time series at the current and previous moments; the other is the difference network which integrates the feature representations of the two input time series with labeled performance at the previous moment to predict the performance at the current moment in an incremental way. Effectiveness of the proposed STS-D net is validated in a real-world froth flotation process.

Deep neural networks are vulnerable to adversarial attacks. More importantly, some adversarial examples crafted against an ensemble of source models transfer to other target models and, thus, pose a security threat to black-box applications (when attackers have no access to the target models). Current transfer-based ensemble attacks, however, only consider a limited number of source models to craft an adversarial example and, thus, obtain poor transferability. Besides, recent query-based black-box attacks, which require numerous queries to the target model, not only come under suspicion by the target model but also cause expensive query cost. In this article, we propose a novel transfer-based black-box attack, dubbed serial-minigroup-ensemble-attack (SMGEA). Concretely, SMGEA first divides a large number of pretrained white-box source models into several ``minigroups.'' For each minigroup, we design three new ensemble strategies to improve the intragroup transferability. Moreover, we propose a new algorithm that recursively accumulates the ``long-term'' gradient memories of the previous minigroup to the subsequent minigroup. This way, the learned adversarial information can be preserved, and the intergroup transferability can be improved. Experiments indicate that SMGEA not only achieves state-of-the-art black-box attack ability over several data sets but also deceives two online black-box saliency prediction systems in real world, i.e., DeepGaze-II (https://deepgaze.bethgelab.org/) and SALICON (http://salicon.net/demo/). Finally, we contribute a new code repository to promote research on adversarial attack and defense over ubiquitous pixel-to-pixel computer vision tasks. We share our code together with the pretrained substitute model zoo at https://github.com/CZHQuality/AAA-Pix2pix.

Accompanied with the rapid increase of the demand for routine examination of leucorrhea, efficiency and accuracy become the primary task. However, in super depth of field (SDoF) system, the problem of automatic detection and localization of cells in leucorrhea micro-images is still a big challenge. The changing of the relative position between the cell center and focus plane of microscope lead to variable cell morphological structure in the two-dimensional image, which is an important reason for the low accuracy of current deep learning target detection algorithms. In this paper, an object detection method based on Retinanet in state of super depth of field is proposed, which can achieve high precision detecting of leucorrhea components by the SDoF feature aggregation module. Compared with the current mainstream algorithms, the mean average accuracy (mAP) index has been improved significantly, the mAP index is 82.7% for SDoF module and 83.0% for SDoF+ module, with an average increase of more than 10%. These improved features can significantly improve the efficiency and accuracy of the algorithm. The algorithm proposed in this paper can be integrated into the leucorrhea automatic detection system.

Motivation A phylogenetic network is a powerful model to represent entangled evolutionary histories with both divergent (speciation) and convergent (e.g. hybridization, reassortment, recombination) evolution. The standard approach to inference of hybridization networks is to (i) reconstruct rooted gene trees and (ii) leverage gene tree discordance for network inference. Recently, we introduced a method called RF-Net for accurate inference of virus reassortment and hybridization networks from input gene trees in the presence of errors commonly found in phylogenetic trees. While RF-Net demonstrated the ability to accurately infer networks with up to four reticulations from erroneous input gene trees, its application was limited by the number of reticulations it could handle in a reasonable amount of time. This limitation is particularly restrictive in the inference of the evolutionary history of segmented RNA viruses such as influenza A virus (IAV), where reassortment is one of the major mechanisms shaping the evolution of these pathogens. Results Here, we expand the functionality of RF-Net that makes it significantly more applicable in practice. Crucially, we introduce a fast extension to RF-Net, called Fast-RF-Net, that can handle large numbers of reticulations without sacrificing accuracy. In addition, we develop automatic stopping criteria to select the appropriate number of reticulations heuristically and implement a feature for RF-Net to output error-corrected input gene trees. We then conduct a comprehensive study of the original method and its novel extensions and confirm their efficacy in practice using extensive simulation and empirical IAV evolutionary analyses. Availability and implementation RF-Net 2 is available at https://github.com/flu-crew/rf-net-2. Supplementary information Supplementary data are available at Bioinformatics online.

A large amount of research has shown the vitality of siphon enumeration in the analysis and control of deadlocks in various resource-allocation systems modeled by Petri nets (PNs). In this paper, we propose an algorithm for the enumeration of minimal siphons in PN based on problem decomposition. The proposed algorithm is an improved version of the global partitioning minimal-siphon enumeration (GPMSE) proposed by Cordone et al. (2005) in IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans, which is widely used in the literature to compute minimal siphons. The experimental results show that the proposed algorithm consumes lower computational time and memory compared with GPMSE, which becomes more evident when the size of the handled net grows.

Object detection is a fundamental and important task in the analysis of remote sensing images (RSIs), and existing deep learning-based object detection models in this literature strongly rely on predefined anchor boxes and encounter redesigned difficulties related to anchors. Additionally, they often ignore the scene-contextual information that objects are usually closely related to their surrounding scene. To deal with these problems, we propose an anchor-free network, referred to as scene-relevant anchor-free network (SRAF-Net), for object detection in RSIs. The SRAF-Net firstly captures the scene-contextual features of objects by using a designed scene-enhanced feature pyramid network (SE-FPN), and then performs more accurate detection by implementing a scene auxiliary detection head (SADH), which can predict the existence of the objects with the help of the scene-contextual features extracted from the SE-FPN. To deal with insufficient scene diversity in the training stage, a simple yet effective data augmentation module, termed balance mixup data augment (BMDA), is introduced by linearly expanding the training dataset to improve the generalization of SRAF-Net. Comprehensive experiments on three publicly-available challenging remote sensing datasets demonstrate the effectiveness of the proposed method. The codes will be made publicly available at https://github.com/Complicateddd/SRAF-Net.

Liver cancer poses a great threat to people's health. Accurate liver segmentation is crucial to the diagnosis of liver cancer. In recent years, great achievements have been made in liver segmentation using a series of improved networks developed based on U‐Net. U‐Net uses a skip connection splices the feature map in the encoder and decoder. However, this method ignores the difference between the two feature maps, which limits the ability of the network to extract liver structures of different sizes. This study proposes a new network structure, the pyramid convolutional attention fusion network (PCAF‐Net), for 2D liver segmentation. The PCAF mechanism was introduced to fuse the feature maps of different receptive field paths after attention mechanism processing to enhance the semantic expression ability of feature maps in skip connection and to improve the segmentation accuracy. The MICCAI 2017 LiTS dataset and CHAOS were used to validate the proposed method. The results of multi‐index evaluation showed that the segmentation performance of the proposed network was superior to those of other networks. PCAF‐Net achieved accurate liver segmentation, providing a reference for artificial intelligenceassisted clinical diagnosis of liver cancer.

Hyperspectral unmixing is recognized as an important tool to learn the constituent materials and corresponding distribution in a scene. The physical spectral mixture model is always important to tackle this problem because of its highly ill-posed nature. In this article, we introduce a linear spectral mixture model (LMM)-based end-to-end deep neural network named SNMF-Net for hyperspectral unmixing. SNMF-Net shares an alternating architecture and benefits from both model-based methods and learning-based methods. On the one hand, SNMF-Net is of high physical interpretability as it is built by unrolling Lp sparsity constrained nonnegative matrix factorization (Lp-NMF) model belonging to LMM families. On the other hand, all the parameters and submodules of SNMF-Net can be seamlessly linked with the alternating optimization algorithm of Lp-NMF and unmixing problem. This enables us to reasonably integrate the prior knowledge on unmixing, the optimization algorithm, and the sparse representation theory into the network for robust learning, so as to improve unmixing. Experimental results on the synthetic and real-world data show the advantages of the proposed SNMF-Net over many state-of-the-art methods.

Skeleton-based action recognition has attracted significant attention and obtained widespread applications due to the robustness of 3D skeleton data. One of the key challenges is how to extract discriminative and robust spatio-temporal features from sparse skeleton data to describe actions and improve recognition accuracy. To address this issue, this paper combines convolutions with attention mechanisms and proposes a deep network for skeleton-based action recognition, termed as local-and-global attention network (LAGA-Net). First, we encode skeleton sequences into joint feature evolution maps to compactly describe the spatial and temporal characteristics of skeleton sequences. Then, a motion guided channel attention module (MGCAM) is proposed to model the interdependencies between feature channels by calculating temporal frame-level motion and enhance motion-salient features in a channel-wise way. Further, a spatio-temporal attention module (STAM) is proposed to model spatio-temporal context-aware collaboration at sequence level and extract spatio-temporal attention features that involve long-range dependencies. Together, MGCAM and STAM are combined to form LAGA-Net, which extracts discriminative features integrating both local and global representations of skeleton sequences. Moreover, a two-stream architecture is proposed to learn complementary features from joint and bone aspects. We conduct extensive experiments to verify the effectiveness and superiority of our proposed method over state-of-the-art approaches on several benchmarks (e.g., NTU RGB+D, Northwestern-UCLA, UTD-MHAD and NTU RGB+D 120).

Given a low-resolution multispectral (MS) image and a high-resolution panchromatic image, the task of pansharpening is to generate a high-resolution MS image. Deep learning (DL)-based methods receive extensive attention recently. Different from the existing DL-based methods, this article proposes a novel deep neural network for pansharpening inspired by the learned iterative soft thresholding algorithm. First, a coupled convolutional sparse coding-based pansharpening (PSCSC) model and related traditional optimization algorithm are proposed. Then, following the procedures of traditional algorithm for solving PSCSC, an interpretable end-to-end deep pansharpening network is developed using a deep unfolding strategy. The designed deep architecture can also be understood in the view of details injection (DI)-based scheme. This work offers a solution that integrates the DL-, DI-, and variational optimization-based schemes into a framework. The experimental results on the reduced- and full-scale datasets demonstrate that the proposed deep pansharpening network outperforms popular traditional methods and some current DL-based methods.

Microwave imaging with large rotating angle and sparse sampling is an attractive approach to obtain the high-resolution target image with reduced radar resource. However, the popular imaging methods, e.g., Range-Doppler (RD), back projection (BP), and sparse recovery (SR), are difficult to deal with large rotating angle and sparse sampling simultaneously. In recent years, deep learning (DL) has been widely studied and been successfully used to handle the problems in computer vision. However, since most existing DL networks are put forward for the real visual image and a large amount of data is essential for network training, DL cannot be directly used to process the complex and sparse target echo for microwave imaging. In this article, a new learning imaging framework is proposed and a model-driven learning imaging network (MDLI-Net) is built for high-resolution microwave imaging with large rotating angle and sparse sampling. In the proposed framework, the electromagnetic scattering model is used to generate the training data efficiently, and the sparse microwave imaging theory is applied to guide the design of the deep imaging network. By inputting the 2-D sparse complex-valued target echo, the trained MDLI-Net can output the high-resolution and focused target image efficiently. The effectiveness of the proposed learning imaging method is validated by experiment results with both simulated and real data.

International audience; In recent years, deep learning (DL) has attracted increasing attention in hyperspectral unmixing (HU) applications due to its powerful learning and data fitting ability. The autoencoder (AE) framework, as an unmixing baseline network, achieves good performance in HU by automatically learning low-dimensional embeddings and reconstructing data. Nevertheless, the conventional AE-based architecture, which focuses more on the pixel-level reconstruction loss, tends to lose some significant detailed information of certain materials (e.g., material-related properties) in the reconstruction process. Therefore, inspired by the perception mechanism, we propose a cycle-consistency unmixing network, called CyCU-Net, by learning two cascaded AEs in an end-to-end fashion, to enhance the unmixing performance more effectively. CyCU-Net is capable of reducing the detailed and material-related information loss in the process of reconstruction by relaxing the original pixel-level reconstruction assumption to cycle consistency dominated by the cascaded AEs. More specifically, cycle consistency can be achieved by a newly proposed self-perception loss, which consists of two spectral reconstruction terms and one abundance reconstruction term. By taking advantage of the self-perception loss in the network, the high-level semantic information can be well preserved in the unmixing process. Moreover, we investigate the performance gain of CyCU-Net with extensive ablation studies. Experimental results on one synthetic and three real hyperspectral data sets demonstrate the effectiveness and competitiveness of the proposed CyCU-Net in comparison with several state-of-the-art unmixing algorithms.

In this letter, we study the problem of non-blockingness verification by tapping into the basis reachability graph (BRG). Non-blockingness is a property that ensures that all pre-specified tasks can be completed, which is a mandatory requirement during the system design stage. We develop a condition of transition partition of a given net such that the corresponding conflict-increase BRG contains sufficient information on verifying non-blockingness of its corresponding Petri net. Thanks to the compactness of the BRG, our approach possesses practical efficiency since the exhaustive enumeration of the state space can be avoided. In particular, our method does not require that the net is deadlock-free.

In hyperspectral image (HSI) classification, it is a great challenge on how to extract key informative spectral-spatial features efficiently and suppress useless features from abundant spectral-spatial information. In this article, inspired by the attention mechanism of the human visual system, we propose a novel spectral-spatial-semantic network (S³Net) with the multiway attention mechanism for HSI classification. The S³Net consists of a spectral branch, a spatial branch, and a multiscale semantic module. The spectral branch extracts the multiway spectral features with a dense spectral block and a multiway spectral attention module. The spatial branch extracts the multiway spatial features with a multiscale spatial block and a multiway spatial attention module. The multiscale semantic module extracts spectral-spatial-semantic features that are used for classification. In the proposed S³Net, the multiway attention modules in spectral and spatial branches are built to enhance the extraction ability of the key informative spectral-spatial features. The multiscale spatial block in the spatial branch is designed to learn strong complementary and related information. The Res2Net in the multiscale semantic module is used to learn multiscale semantic features at a granular level. A large number of experimental results demonstrate that, on the University of Pavia, Kennedy Space Center, and Pavia Center data sets, the proposed S³Net achieves higher classification accuracy than state-of-the-art methods on the limited training samples. Remarkably, our S³Net also achieves the best performance on the mineral exploration HSI data set, called Huoshaoyun, which is collected by the GaoFen-5 (GF-5) satellite.

Recently, the development of 3d depth technology brought in many real-world multimedia applications, however, how to tackle real-world degradations in low-resolution (LR) depth maps remains a major challenge, though remarkable progresses have been achieved with DCNN based depth super-resolution (DSR) approaches. Existing DSR models are usually trained and tested on synthetic dataset, which are restrictive and not effective in generalizing to the real-world DSR tasks. In this paper, we aim at alleviating the real-world degradations of different depth sensors in two aspects. First, we classify the generation of LR depth maps into two types: non-linear down-sampling with noise and interval down-sampling, for which different DSR models are learned correspondingly. Second, a novel framework is proposed to handle these two types of LR depth maps in DSR, which consists of four modules: 1) An progressive residual learning module with deep supervision is proposed to learn effective high-frequency components of depth maps in a coarse-to-fine manner. 2) A weighted attention fusion strategy is utilized to intensify the features with abundant high-frequency components in both global and local manners. 3) A multi-stage fusion module is utilized to sufficiently re-exploit the information in the progressive process. 4) A depth refinement module is proposed to improve the depth map by Total Generalized Variation (TGV) regularization and input loss. Extensive experiments on benchmarking datasets demonstrate the superiority of our method over current state-of-the-art DSR methods.

A rumor about an entity can spread rapidly through online social networks (OSNs), which may lead to serious consequences. This article focuses on developing a cost-effective rumor-refuting strategy in the situation where the rumormonger is strategic. Based on a novel individual-based rumor spreading model, we estimate the expected net benefit of the rumormonger as well as the expected total loss of the rumor victim. On this basis, we reduce the original problem to a differential game-theoretic model. Next, we derive a system for solving the model. By solving the system, we get a dynamic strategy pair. Through extensive comparative experiments, we find that this strategy pair is effective in terms of the solution concept of Nash equilibrium. Hence, we conclude that the rumor-refuting strategy contained in this strategy pair is cost-effective. Finally, we examine the effect of the structure of the OSN on the cost effectiveness of the rumor-supporting and rumor-refuting strategies contained in the proposed strategy pair. To our knowledge, this is the first time the rumor-refuting problem is addressed under the assumption that the rumormonger is strategic.

We propose the "Tile-net" method based on dividing an image into smaller tiles. Using the tile as the input to the neural network, the network is simplified substantially. The Tile-net learns at a much faster rate than the networks without tiling. The training and reconstruction times for the Tile-net are reduced by 40% and 33%, respectively compared to the networks without tiling. The Tile-net performance is evaluated through the normalized mean square error (NMSE), peak signal to noise ratio (PSNR), structure similarity index measure (SSIM) and the quality of the reconstructed image for test datasets. The Tile-net does not degrade performance; however, it reduces the NMSE by 0.3% compared to the networks without tiling.

Convolutional neural networks (CNNs) are prone to mis-segmenting image data of the liver when the background is complicated, which results in low segmentation accuracy and unsuitable results for clinical use. To address this shortcoming, an interactive liver segmentation algorithm based on geodesic distance and V-net is proposed. The three-dimensional segmentation network V-net adequately considers the characteristics of the spatial context information to segment liver medical images and obtain preliminary segmentation results. An artificial algorithm based on geodesic distance is used to form artificial hard constraints to modify the image, and the superpixel piece created by the watershed algorithm is introduced as a sample point for operation, which significantly improves the efficiency of segmentation. Results from simulation of the liver tumor segmentation challenge (LiTS) dataset show that this algorithm can effectively refine the results of automatic liver segmentation, reduce user intervention, and enable a fast, interactive liver image segmentation that is convenient for doctors.

The appearance-based method for gaze estimation has great potential to work well under various conditions, but current learning-based methods ignore inferior eye images in datasets caused by poor eye region locating, occlusions and abnormal head poses. These images badly impact the accuracy of estimation. In the study, inspired by binocular vision characteristics, we propose two cooperative sub-networks, True Gaze Consistency Network (TG-Net) and Single Gaze Inconsistency Network(SG-Net) which composes TSG-Net. TG-Net and SG-Net cooperate through a residual paradigm and Informing module. More specially, TG-Net explicitly extracts the consistency of paired eyes and weights high-level features from two paired-eye images utilizing SE-Block and an artificial gaze direction, named True Gaze. SG-Net outputs residual momentums based on True Gaze for better estimation of paired eyes. Experimental results on three benchmark datasets demonstrate that the proposed method performs competitively against the existed representative CNN-Based methods. TSG-Net improves on the state-of-the-art by 22% on MPIIGaze and shows more advantages in additional analysis.

Most existing segmentation networks are built upon a “ U -shaped” encoder–decoder structure, where the multi-level features extracted by the encoder are gradually aggregated by the decoder. Although this structure has been proven to be effective in improving segmentation performance, there are two main drawbacks. On the one hand, the introduction of low-level features brings a significant increase in calculations without an obvious performance gain. On the other hand, general strategies of feature aggregation such as addition and concatenation fuse features without considering the usefulness of each feature vector, which mixes the useful information with massive noises. In this article, we abandon the traditional “ U -shaped” architecture and propose Y-Net, a dual-branch joint network for accurate semantic segmentation. Specifically, it only aggregates the high-level features with low-resolution and utilizes the global context guidance generated by the first branch to refine the second branch. The dual branches are effectively connected through a Semantic Enhancing Module, which can be regarded as the combination of spatial attention and channel attention. We also design a novel Channel-Selective Decoder (CSD) to adaptively integrate features from different receptive fields by assigning specific channelwise weights, where the weights are input-dependent. Our Y-Net is capable of breaking through the limit of singe-branch network and attaining higher performance with less computational cost than “ U -shaped” structure. The proposed CSD can better integrate useful information and suppress interference noises. Comprehensive experiments are carried out on three public datasets to evaluate the effectiveness of our method. Eventually, our Y-Net achieves state-of-the-art performance on PASCAL VOC 2012, PASCAL Person-Part, and ADE20K dataset without pre-training on extra datasets.

This study explores RAU-Net convolutional network architecture on satellite images by classifying geospatial objects on satellite images such as roofs to help in determining the building density of...

A traditional image steganography embeds secret information into a cover image to generate a secret-embedded image. The modification traces imposed on the cover image can be easily detected by steganalysis tools. Coverless steganography has been introduced to solve this problem. In this study, coverless steganography is combined with image style transfer, an arbitrary image style transfer network CSST-Net is put forward, and a secret information is encoded into the parameters (an adaptive steganography matrix) of CSST-Net, which is used to restrict the style transfer. Arbitrary image style transfer is performed instructed by the adaptive steganography matrix, and the image style transfer result driven by secret information is directly synthesized. Our experiments show that CSST-Net can not only synthesize any image style transfer result with good visual effect, but also achieve good performance in capacity, anti-steganalysis and security.

The recent success of deep convolutional neural networks (CNNs) is mostly due to the availability of large-scale datasets with accurate annotations. However, the collection of such large datasets with clean annotations is time-consuming and not always feasible. In this paper, we propose a novel framework for learning with noisy labels, called the Cyclical training method with Joint loss and Co-teaching strategy net (CJC-net), where the net means our method is insensitive to the structures of the CNN. CJC-net pretrains two networks simultaneously and then performs the cyclical training strategy under an improved co-teaching method based on the two pretrained networks. During the training process, we adjust the learning rates of the two networks to make the network states periodically transfer from overfitting to underfitting. The cumulative loss of each sample under two networks is recorded; and the higher the cumulative loss of a sample is, the higher the probability of identifying it as a noisy label or a hard label. Then, we remove those samples with a high loss and fine-tune the two networks using the remaining data. The experimental results on several datasets demonstrate that CJC-net is superior to many state-of-the-art methods.

Vehicle segmentation, traffic density estimation and vehicle tracking are some of the important functionalities in smart traffic management. Segmentation of vehicles from traffic surveillance videos can help to realize many applications like speed monitoring, traffic estimation etc. Vehicle segmentation from surveillance videos becomes very challenging due to presence of occlusions, cluttered backgrounds and traffic density variations. In this work, we propose a deep learning adapted active net segmentation model for vehicle segmentation. The proposed solution involves three stages: adaptive background model based subtraction, active net sub netting with CNN, refined from CNN results with ETAN optimization. Adaptive background modeling is based on an adaptive gain function which is constructed from the pixels of the frames in the video. The gain function is able to compensate for the shadow and illumination problems affecting the vehicle segmentation. Deep learning assisted topological active net deformable model is able to provide higher segmentation accuracy in presence of occlusions, cluttered backgrounds and traffic density variations.

MRI image analysis and its segmentation for the accurate and automatic detection of brain tumors at an early stage is very much crucial for diagnosis the disorders and save human lives. Since most deep learning models have a large number of layers, they also take longer processing time, making them unsuitable for smaller image datasets. Hence, we have proposed, the detection of abnormality from brain MR images using a Less Layered and less complex U-Net model (LeU-Net) architecture. The principle of LeU-Net is inspired by the Le-Net and U-Net models, but completely different from both the design and architectural perspectives. The abnormality detection indicates the classification of the tumorous cell from overall Magnetic Resonance images. The Proposed deep learning model (LeU-Net) performance was compared with the existing basic CNN models Le-Net, U-Net, and VGG-16. The model performance was evaluated using evaluation metrics accuracy, precision, F-score, recall, and specificity. The experiment is performed on MR Dataset with uncropped images and cropped images (removed unwanted area) and compared the result with all three models. The LeU-Net model registers overall 98% accuracy on cropped images and 94% of accuracy on uncropped images. The LeU-Net model has much faster processing (simulation) time, it only takes 244.42 s and 252.36 s, respectively, to train the model with 100 epochs on the uncropped and cropped images. We have compared the performance of our proposed model with various state-of-the-art techniques, and it provides the best classification accuracy among all.

As obesity becomes increasingly common worldwide [1], more people want to lose weight to improve their health and image. According to the Centers for Disease Control and Prevention (CDC), long-term changes in daily eating habits (such as regarding food/ nutrition type, calorie intake) are successful at keeping weights off [2]. Therefore, it would be helpful to have an artificial intelligence (AI) mobile program that identifies the types of food the user consumes and automatically calculates the total calories. This paper examines the development and optimization of an 11-categorical food classification model based on the Mobile-Net neural network using Python. Specifically, it classifies any food image as one of bread, dairy, dessert, egg product, fried food, meat, noodles, rice, seafood, soup, or fruit/vegetables. Methods of optimization include data preprocessing and learning rate and batch size adjustments. Experimental results show that scaling image inputs to standard size (Python Numpy resize) function), 300 training epochs, dynamic learning rate (start with 0.001 and *0.1 for every 30 epochs), and a batch size of 16 yields our best model of 83.44% accuracy.

We develop a randomized algorithm (that succeeds with high probability) for generating an $$\epsilon $$ -net in a sphere of dimension n. The basic scheme is to pick an alphabet consisting of $$O (n\ln (1/\epsilon )+\ln (1/\delta ))$$ random rotations, form all possible words of length $$O (n\ln (1/\epsilon ))$$ from this alphabet, and require these words act on a fixed point. We show the set of points so generated is equidistributed at a scale of $$\epsilon $$ .

In the present paper we prove a weak form of sandwich classification for the overgroups of the subsystem subgroup $E(\Delta,R)$ of the Chevalley group $G(\Phi,R)$ where $\Phi$ is a simply laced root sysetem and $\Delta$ is its sufficiently large subsystem. Namely we show that for any such an overgroup $H$ there exists a unique net of ideals $\sigma$ of the ring $R$ such that $E(\Phi,\Delta,R,\sigma)\le H\le {\mathop{\mathrm{Stab}}\nolimits}_{G(\Phi,R)}(L(\sigma))$ where $E(\Phi,\Delta,R,\sigma)$ is an elementary subgroup associated with the net and $L(\sigma)$ is a corresponding subalgebra of the Chevalley Lie algebra., Comment: 25 pages, to appear in St.Petersburg Math Journal

This article provides a systematic review of the literature on net-zero carbon cities, their objectives and key features, current efforts, and performance. We discuss how net-zero differs from low-carbon cities, how different visions of a net-zero carbon city relate to urban greenhouse gas accounting, deep decarbonization pathways and their application to cities and urban infrastructure systems, net-zero carbon cities in theory versus practice, lessons learned from net-zero carbon city plans and implementation, and opportunities and challenges in transitioning toward net-zero carbon cities across both sectors and various spatial fabrics within cities. We conclude that it is possible for cities to get to or near net-zero carbon, but this requires systemictransformation. Crucially, a city cannot achieve net-zero by focusing only on reducing emissions within its administrative boundaries. Cities must decarbonize key transboundary supply chains and use urban and regional landscapes to sequester carbon from the atmosphere. Because of carbon lock-in, and the complex interplay between urban infrastructure and behavior, strategic sequencing of mitigation action is essential for cities to achieve net-zero.

With pedestrian detection algorithms, balancing the trade-off between accuracy and speed remains challenging. Following the central point-based one-stage object detection paradigm, a pedestrian detection algorithm based on multi-scale attention feature aggregation (MAFA) is proposed to improve accuracy while considering real-time performance. We refer to the proposed algorithm as MAFA-Net. Through the design of deep dilate blocks, deeper features are extracted. Pedestrian attention blocks are added to mine more relevant information between features from the perspective of spatial and passage-wise dimensions, and pedestrian features are enhanced. Feature aggregation modules are used to fuse different scale features, and combine the rich high-level semantic features with the accurate location features of the low-level features. Experiments were conducted on two challenging pedestrian detection datasets, i.e., CityPersons and Caltech, using MR−2 as the evaluation index. For Caltech, MR−2 is 4.58% under reasonable conditions. For CityPersons, MR−2 is 11.47% and 10.05% under reasonable and partial occlusion conditions, which is 0.43% and 1.35% better than the suboptimal comparison detection method. The results demonstrate that a good performance is obtained, and the effectiveness and feasibility of the algorithm are verified.

Language keyword extraction has broad application prospects. Most current keyword search networks are based on speech recognition and text search. This method is helpless for some languages that are difficult to text and cannot locate the location where the keywords appear. A multi‐level structure of speech keyword location and recognition system (MSS‐Net), which can locate the location interval of keywords while recognizing keywords is proposed here. MSS‐Net contains three levels. The first level is the voice start and end detection module, which is responsible for detecting the start and end positions of valid audio. When valid audio is detected, the second level is activated: the keyword detection module. The keyword detection module detects keyword information in the audio signal and locates the keyword position interval. The third level is the keyword recognition module. The system recognizes the detected position interval of the location to obtain specific keyword information. Comparative experiments show that MSS‐Net is more accurate for keyword positioning, with a network recall rate of 94.91% and an accuracy rate of 96.32%, which is better than other network models participating in the experiment. In addition, the network has strong anti‐interference ability, which can achieve good suppression effect for normal life noise. The ablation experiment verified the effect of the hierarchical network structure on the performance of speech recognition tasks.

Glaucoma is a disease caused by fluid pressure build-up in the inner eye. Early detection of glaucoma is critical as it is expected that 111.8 million people worldwide shall suffer from glaucoma in 2040. In the diagnosis of glaucoma, the use of machine learning method is hoped to be highly promising. This paper provides an important method to master learning to diagnose glaucoma. Initially, human retinal fundus images are preprocessed by means of histogram equalization in order to enhance them. The segmentation is performed by semantic segmentation method, mainly the features are extracted using density with correlation based feature extraction approach. PCA (principal component analysis) methodology is used to choose the most optimal features. Ultimately, through the usage of the Deep residual Google Net CNN Classification method, the retinal image is classified/predicted as regular and abnormal. The Deep residual Google Net CNN classifier is designed to distinguish view patterns with minimal pre-processing from pixel pictures. ORIGA and STARE datasets are used in this work. The findings are then analyzed and contrasted to illustrate the efficacy of the new technique with alternate current techniques. Test accuracy of 99%, Specificity of 98.9% and 100% Sensitivity were achieved. The quantitative results are analyzed for specifications like sensitivity, specificity, accuracy, positive predictive rate, false predictive rate and assured to provide excellent outcomes when compared with traditional methods.

Steganalysis is the technique that tries to beat steganography by detecting and removing secret information. Steganalysis involves the detection of a message embedded in a picture. Deep Learning (DL) advances have offered alternative approaches to many difficult issues, including the field of image steganalysis using deep-learning architecture based on convolutionary neural networks (CNN). In recent years, many CNN architectures have been established that have enhanced the exact identification of steganographic images. This work presents a novel architecture which involves a preprocessing stage using histogram equalization and adaptive recursive median filter banks to reduce image noise, a feature extraction stage using shearlet multilinear local embedding methods and then finally the classification can be done by using the discrete scalable Alex NET CNN classifier. Performance was evaluated on the RGB-BMP Steganalysis Dataset with different experimental setups. To prove the effectiveness of the suggested algorithm it can be compared with the other existing methodologies. This work improves classification accuracies on all other existing algorithms over test data.

According to Löwenstein's rule, Al–O–Al bridges are forbidden in the aluminosilicate framework of zeolites. A graph-theoretical interpretation of the rule, based on the concept of independent sets, was proposed earlier. It was shown that one can apply the vector method to the associated periodic net and define a maximal Al/(Al+Si) ratio for any aluminosilicate framework following the rule; this ratio was called the independence ratio of the net. According to this method, the determination of the independence ratio of a periodic net requires finding a subgroup of the translation group of the net for which the quotient graph and a fundamental transversal have the same independence ratio. This article and a companion paper deal with practical issues regarding the calculation of the independence ratio of mainly 2-periodic nets and the determination of site distributions realizing this ratio. The first paper describes a calculation technique based on propositional calculus and introduces a multivariate polynomial, called the independence polynomial. This polynomial can be calculated in an automatic way and provides the list of all maximal independent sets of the graph, hence also the value of its independence ratio. Some properties of this polynomial are discussed; the independence polynomials of some simple graphs, such as short paths or cycles, are determined as examples of calculation techniques. The method is also applied to the determination of the independence ratio of the 2-periodic net dhc.

Real-time segmentation and understanding of driving scenes are crucial in autonomous driving. Traditional pixel-wise approaches extract scene information by segmenting all pixels in a frame, and hence are inefficient and slow. Proposal-wise approaches only learn from the proposed object candidates, but still require multiple steps on the expensive proposal methods. Instead, this work presents a fast single-shot segmentation strategy for video scene understanding. The proposed net, called S3-Net, quickly locates and segments target sub-scenes , and meanwhile extracts attention-aware time-series sub-scene features ( ats-features ) as inputs to an attention-aware spatio-temporal model (ASM) . Utilizing tensorization and quantization techniques, S3-Net is intended to be lightweight for edge computing. Experiments results on CityScapes, UCF11, HMDB51, and MOMENTS datasets demonstrate that the proposed S3-Net achieves an accuracy improvement of 8.1% versus the 3D-CNN based approach on UCF11, a storage reduction of 6.9× and an inference speed of 22.8 FPS on CityScapes with a GTX1080Ti GPU.