Search

The segmentation of neuronal morphology in electron microscopy images is crucial for the analysis and understanding of neuronal function. However, most of the existing segmentation methods are not suitable for challenging datasets where the neuronal structure is contaminated by noise or has interrupted parts. In this paper, we propose a segmentation method based on deep learning to determine the location information of neurons and reduce the influence of image noise in the data. Specifically, we adapt our neuron dataset based on UNet by using convolution with BN fusion and multi-input feature fusion. The method is named REDAFNet. The model simplifies the model structure and enhances the generalization ability by fusing the convolution layer and BN layer. The noise interference in the data was reduced by multi-input feature fusion, and the ability to understand and express the data was enhanced. The method takes a neuron image as input and its pixel segmentation map as output. Experimental results show that the segmentation accuracy of the proposed method is 91.96%, 93.86% and 80.25% on the ISBI2012 dataset, U-RISC retinal neuron dataset and N2DH-GOWT1 stem cell dataset, respectively. Compared with the existing segmentation methods, the proposed method can extract more complete feature information and achieve more accurate segmentation. [ABSTRACT FROM AUTHOR]

With the development of artificial intelligence, deep-learning-based log anomaly detection proves to be an important research topic. In this paper, we propose LogCSS, a novel log anomaly detection framework based on the Context-Semantics-Statistics Convolutional Neural Network (CSSCNN). It is the first model that uses BERT (Bidirectional Encoder Representation from Transformers) and CNN (Convolutional Neural Network) to extract the semantic, temporal, and correlational features of the logs. We combine the features with the statistic information of log templates for the classification model to improve the accuracy. We also propose a technique, DOOT (Deals with the Out-Of-Templates), for online template matching. The experimental research shows that our framework improves the average F1 score of the six best algorithms in the industry by more than 5% on the open-source dataset HDFS, and improves the average F1 score of the six best algorithms in the industry by more than 8% on the BGL dataset, LogCSS also performs better than other similar methods on our own constructed dataset. [ABSTRACT FROM AUTHOR]

Military object military object detection technology serves as the foundation and critical component for reconnaissance and command decision-making, playing a significant role in information-based and intelligent warfare. However, many existing military object detection models focus on exploring deeper and more complex architectures, which results in models with a large number of parameters. This makes them unsuitable for inference on mobile or resource-constrained combat equipment, such as combat helmets and reconnaissance Unmanned Aerial Vehicles (UAVs). To tackle this problem, this paper proposes a lightweight detection framework. A CSP-GhostnetV2 module is proposed in our method to make the feature extraction network more lightweight while extracting more effective information. Furthermore, to fuse multiscale information in low-computational scenarios, GSConv and the proposed CSP-RepGhost are used to form a lightweight feature aggregation network. The experimental results demonstrate that our proposed lightweight model has significant advantages in detection accuracy and efficiency compared to other detection algorithms. [ABSTRACT FROM AUTHOR]

As a list of remotely sensed data sources is available, the effective processing of remote sensing images is of great significance in practical applications in various fields. This paper proposes a new lightweight network to solve the problem of remote sensing image processing by using the method of deep learning. Specifically, the proposed model employs ShuffleNet V2 as the backbone network, appropriately increases part of the convolution kernels to improve the classification accuracy of the network, and uses the maximum overlapping pooling layer to enhance the detailed features of the input images. Finally, Squeeze and Excitation (SE) blocks are introduced as the attention mechanism to improve the architecture of the network. Experimental results based on several multisource data show that our proposed network model has a good classification effect on the test samples and can achieve more excellent classification performance than some existing methods, with an accuracy of 91%, and can be used for the classification of remote sensing images. Our model not only has high accuracy but also has faster training speed compared with large networks and can greatly reduce computation costs. The demo code of our proposed method will be available at https://github.com/li-zi-qi. [ABSTRACT FROM AUTHOR]

The Central Nervous System (CNS) is one of the most crucial parts of the human body. Brain tumor is one of the deadliest diseases that affect CNS and they should be detected earlier to avoid serious health implications. As it is one of the most dangerous types of cancer, its diagnosis is a crucial part of the healthcare sector. A brain tumor can be malignant or benign and its grade recognition is a tedious task for the radiologist. In the recent past, researchers have proposed various automatic detection and classification techniques that use different imaging modalities focusing on increased accuracy. In this paper, we have done an in-depth study of 19 different trained deep learning models like Alexnet, VGGnet, DarkNet, DenseNet, ResNet, InceptionNet, ShuffleNet, NasNet and their variants for the detection of brain tumors using deep transfer learning. The performance parameters show that NASNet-Large is outperforming others with an accuracy of 98.03% for detection and 97.87% for classification. The thresholding algorithm is used for segmenting out the tumor region if the detected output is other than normal. [ABSTRACT FROM AUTHOR]

BACKGROUND: An effective method for achieving low-dose CT is to keep the number of projection angles constant while reducing radiation dose at each angle. However, this leads to high-intensity noise in the reconstructed image, adversely affecting subsequent image processing, analysis, and diagnosis. OBJECTIVE: This paper proposes a novel Channel Graph Perception based U-shaped Transformer (CGP-Uformer) network, aiming to achieve high-performance denoising of low-dose CT images. METHODS: The network consists of convolutional feed-forward Transformer (ConvF-Transformer) blocks, a channel graph perception block (CGPB), and spatial cross-attention (SC-Attention) blocks. The ConvF-Transformer blocks enhance the ability of feature representation and information transmission through the CNN-based feed-forward network. The CGPB introduces Graph Convolutional Network (GCN) for Channel-to-Channel feature extraction, promoting the propagation of information across distinct channels and enabling inter-channel information interchange. The SC-Attention blocks reduce the semantic difference in feature fusion between the encoder and decoder by computing spatial cross-attention. RESULTS: By applying CGP-Uformer to process the 2016 NIH AAPM-Mayo LDCT challenge dataset, experiments show that the peak signal-to-noise ratio value is 35.56 and the structural similarity value is 0.9221. CONCLUSIONS: Compared to the other four representative denoising networks currently, this new network demonstrates superior denoising performance and better preservation of image details. [ABSTRACT FROM AUTHOR]

Osteosarcomas are a type of bone tumour that can develop anywhere in the bone but most typically do so around the metaphyseal growth plates at the ends of long bones. Death rates can be lowered by early detection. Manual osteosarcoma identification can be difficult and requires specialised knowledge. With the aid of contemporary technology, medical photographs may now be automatically analysed and categorised, enabling quicker and more effective data processing. This paper proposes a novel hyperparameter-tuned deep learning (DL) approach for predicting osteosarcoma on histology images with effective feature selection mechanism which aims to improve the prediction accuracy of the classification system for bone tumor detection. The proposed system mainly consists of '6' phases: data collection, preprocessing, segmentation, feature extraction, feature selection, and classification. Firstly, the dataset of histology images is gathered from openly available sources. Then Median Filtering (MEF) is utilized as the preprocessing step that enhances the quality of the input images for accurate prediction by eliminating unwanted information from them. Afterwards, the pre-processed image was segmented using Harmonic Mean-based Otsu Thresholding (HMOTH) approach to obtain the tumor-affected regions from the pre-processed data. Then the features from the segmented tumor portions are extracted using the Self-Attention Mechanism-based MobileNet (SAMMNet) model. A Van der Corput sequence and Adaptive Inertia Weight included Reptile Search Optimization Algorithm (VARSOA) is used to select the more relevant features from the extracted features. Finally, a Hyperparameter-Tuned Deep Elman Neural Network (HTDENN) is utilized to diagnose and classify osteosarcoma, in which the hyperparameters of the neural network are obtained optimally using the VARSOA. The proposed HTDENN attains the higher accuracy of 0.9531 for the maximum of 200 epochs, whereas the existing DENN, MLP, RF, and SVM attains the accuracies of 0.9492, 0.9427, 0.9413, and 0.9387. Likewise, the proposed model attains the better results for precision (0.9511), f-measure (0.9423), sensitivity (0.9345) and specificity (0.9711) than the existing approaches for the maximum of 200 epochs. Simulation outcomes proved that the proposed model outperforms existing research frameworks for osteosarcoma prediction and classification. [ABSTRACT FROM AUTHOR]

With the rapid growth of Internet video image information, there is a large amount of redundancy in image data. Use less data stream information to transfer the image or the amount of information contained in the image. Its purpose is to reduce the redundancy of images, so as to store them at low bit rate and reduce the data storage space. In the general image compression method, the hybrid coding framework is adopted. Each algorithm adopts a fixed algorithm mode through a specific design algorithm, without global optimization. Image compression is mainly divided into prediction, transformation, quantization, digital entropy coding and other steps. At present, there are many researches on super-resolution network based on deep learning technology. The main function is to reconstruct high-resolution image replace image magnification low-resolution images such as linear interpolation, which has a great performance improvement image resolution, noise reduction, deblurring and so on, but there is no effective way to use super-resolution network applications to improve quality of compression reconstructed image quality. This paper involves a new method that using image super-resolution residual learning network to improve quality of compression image, our method, the reduced image is encoded into a content stream and a transmission corresponding parameter is encoded into a model stream. Firstly, the original image is scaled down 1/2 size of source image, then encode the small image into content stream with the existing codec. Secondly, the residual learning super-resolution (SR) network is used for image filtering to scale up reconstructed image with decode image resizing method and boost the quality of edge feature extraction of image. Our results show that there is significant performance improvement of h265 in low resolution reconstructed image (bits-per-pixel less than 0.1). [ABSTRACT FROM AUTHOR]

Human posture detection is easily affected by the external environment, resulting in blurred results of limb feature extraction. In order to improve the accuracy and speed of human motion detection, this paper proposes a deep learning-based motion detection method in competitive sports training. The double parallel convolution network algorithm in the depth learning algorithm is used to process the collected action information, extract the body action features, and greatly reduce the operation scale; With the help of the theory of motion mechanics, the mechanical parameters in the motion process are calculated to eliminate outliers and reduce feature dimensions; With the help of motion inertial sensors and joint degrees of freedom, the limb motion detection results are obtained. The experimental results show that the average recognition rate of the method for different motion actions is 99.5%, and the average detection time is 148 ms, with good application performance. [ABSTRACT FROM AUTHOR]

In the field of image classification, the Convolutional Neural Networks (CNNs) are effective. Most of the work focuses on improving and innovating CNN's network structure. However, using labeled data more effectively for training has also been an essential part of CNN's research. Combining image disturbance and consistency regularization theory, this paper proposes a model training method (PairTraining) that takes image pairs as input and dynamically modify the training difficulty according to the accuracy of the model in the training set. According to the accuracy of the model in the training set, the training process will be divided into three stages: the qualitative stage, the fine learning stage and the strengthening learning stage. Contrastive learning images are formed using a progressively enhanced image disturbance strategy at different training stages. The input image and contrast learning image are combined into image pairs for model training. The experiments are tested on four public datasets using eleven CNN models. These models have different degrees of improvement in accuracy on the four datasets. PairTraining can adapt to a variety of CNN models for image classification training. This method can better improve the effectiveness of training and improve the degree of generalization of classification models after training. The classification model obtained by PairTraining has better performance in practical application. [ABSTRACT FROM AUTHOR]

Robust and reliable 3D multi-object tracking (MOT) is essential for autonomous driving in crowded urban road scenes. In those scenarios, accurate data association between tracked objects and incoming new detections is crucial. This paper presents a tracking system based on the Kalman filter that uses a deep learning approach to the association problem. The proposed architecture consists of three neural networks. First, a convolutional LSTM network extracts spatiotemporal features from a sequence of detections of the same track. Then, a Siamese network calculates the degree of similarity between all tracks and the new detections found at each new frame. Finally, a recurrent LSTM network is used to extract 3D and bounding box information. This model follows the tracking-by-detection paradigm and has been trained with track sequences to be able to handle missed observations and to reduce identity switches. A validation test was carried out on the Argoverse dataset to validate the performance of the proposed system. The developed deep learning approach could improve current multi-object tracking systems based on classic algorithms like the Kalman filter. [ABSTRACT FROM AUTHOR]

The task of cell segmentation in microscope images is difficult and popular. In recent years, deep learning-based techniques have made incredible progress in medical and microscopy image segmentation applications. In this paper, we propose a novel deep learning approach called Residual-Atrous MultiResUnet with Channel Attention Mechanism (RAMRU-CAM) for cell segmentation, which combines MultiResUnet architecture with Channel Attention Mechanism (CAM) and Residual-Atrous connections. The Residual-Atrous path mitigates the semantic gap between the encoder and decoder stages and manages the spatial dimension of feature maps. Furthermore, the Channel Attention Mechanism (CAM) blocks are used in the decoder stages to better maintain the spatial details before concatenating the feature maps from the encoder phases to the decoder phases. We evaluated our proposed model on the PhC-C2DH-U373 and Fluo-N2DH-GOWT1 datasets. The experimental results show that our proposed model outperforms recent variants of the U-Net model and the state-of-the-art approaches. We have demonstrated how our model can segment cells precisely while using fewer parameters and low computational complexity. [ABSTRACT FROM AUTHOR]

Handwritten signature for the identification and authentication of an individual has been widely used in the biometric systems. Due to the intra-class and inter-class variabilities, signature verification has become one of the most challenging problem in the biometric technology. Furthermore, the offline handwritten signature can be forged by the skilled persons due to its static nature. Therefore, in this paper a deep learning-based method using convolutional neural network (CNN) for online signature verification has been developed. Different values of the convolutional kernels such as 1×1, 3×3 and 5×5 are used to extract the discriminative features at multi-scales. The features of the initial and middle layers of the CNN are combined to create more powerful features. An up-sampling method with bilinear interpolation has been used to add the features of convolutional layers with different spatial dimensions. Both the addition and concatenation methods have been used to aggregate the convolutional features. A convolutional transpose method is applied to increase the depth of the convolutional layers while performing an addition operation on the layers with different depths. Finally, the concatenated features are passed to the fully connected layers for high-level feature extraction and classification. To evaluate the performance of the proposed method, an android application was developed where; a custom database of 985 online signatures collected from 197 users has been created. The problem of inadequate training data for online signature verification has been addressed through the data augmentation method. The experimental results show that the deep aggregated convolutional feature representation method achieves an accuracy of 99.32% on the custom developed online signature database. [ABSTRACT FROM AUTHOR]

Because of the advantages of deep learning and information fusion technology, it has drawn much attention for researchers to combine them to achieve target recognition, positioning, and tracking. However, when the existing neural network process multichannel images (e.g., color images), multiple channels as a whole input into neural networks, which makes it hard for networks to fully learn information in R, G, and B channels of images. Therefore, it is not conducive to the final learning effect of the networks. To solve the problem, using different combinations of R, G, and B channels of color images for feature-level fusion, this paper proposes three fusion types as "R/G/B", "R+G/G+B/B+R", and "R+G+B/R+G+B/R+G+B" multichannel concat-fusional convolutional neural networks. Experimental results show that multichannel concat-fusional convolutional neural networks with fusional types of "R+G/G+B/B+R" and "R+G+B/R+G+B/R+G+B" achieve better performance than the corresponding non-fusional convolutional neural networks on different datasets. It shows that networks with fusion types of "R+G/G+B/B+R" and "R+G+B/R+G+B/R+G+B" can learn more fully information of R, G, and B channels of color images and improve the learning performance of networks. [ABSTRACT FROM AUTHOR]

Hyperspectral Image (HSI) is usually composed of hundreds of capturing wavelength bands, which not only increase the size of the HSI rapidly but also impose various obstacles in classifying the objects accurately. Moreover, the traditional machine learning schemes utilize only the spectral features for HSI classification, which, therefore, neglect the spatial features that have a significant impact on the classification improvement. To address the aforementioned issues, in this paper, we propose to employ the principal component analysis (PCA), the baseline feature extraction method, and a thoughtfully designed stacked autoencoder, a deep learning-based feature extraction approach, for reducing the high dimensionality of the HSI and then propose a novel lightweight 3D-2D convolutional neural network (CNN) framework to concurrently exploit both spatial and spectral features from the dimensionality-reduced HSI for classification. In particular, PCA and stacked autoencoder are applied to reduce the high dimensionality of the original HSI and then the proposed 3D-2D CNN provides a combination of 3D and 2D convolution operations to extract the subtle spatial and spectral features for efficient classification. We well-adjust the proposed 3D-2D CNN architecture, and perform extensive experiments on three benchmark HSI datasets and compare our approach with the state-of-the-art classical and deep learning methods. Experimental results illustrate that we have achieved an overall accuracy of 99.73%, 99.90%, and 99.32% on Indian Pines, Pavia University, and Kennedy Space Center datasets, respectively, which outperform the classical machine learning and independent 2D and 3D CNN-based state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Malaria is one of the three major diseases with the highest mortality worldwide and can turn fatal if not taken seriously. The key to surviving this disease is its early diagnosis. However, manual diagnosis is time consuming and tedious due to the large amount of image data. Generally, computer-aided diagnosis can effectively improve doctors' perception and accuracy. This paper presents a medical diagnosis method powered by convolutional neural network (CNN) to extract features from images and improve early detection of malaria. The image sharpening and histogram equalization method are used aiming at enlarging the difference between parasitized regions and other area. Dropout technology is employed in every convolutional layer to reduce overfitting in the network, which is proved to be effective. The proposed CNN model achieves a significant performance with the best classification accuracy of 99.98%. Moreover, this paper compares the proposed model with the pretrained CNNs and other traditional algorithms. The results indicate the proposed model can achieve state-of-the-art performance from multiple metrics. In general, the novelty of this work is the reduction of the CNN structure to only five layers, thereby greatly reducing the running time and the number of parameters, which is demonstrated in the experiments. Furthermore, the proposed model can assist clinicians to accurately diagnose the malaria disease. [ABSTRACT FROM AUTHOR]

In a real-time application such as traffic monitoring, it is required to process the enormous amount of data. Traffic prediction is essential for intelligent transportation systems (ITSs), traffic management authorities, and travelers. Traffic prediction has become a challenging task due to various non-linear temporal dynamics at different locations, complicated underlying spatial dependencies, and more extended step forecasting. To accommodate these instances, efficient visualization and data mining techniques are required to predict and analyze the massive amount of traffic big data. This paper presents a deep learning-based parallel convolutional neural network (Parallel-CNN) methodology to predict the traffic conditions of a specific region. The methodology of deep learning contains multiple processing layers and performs various computational strategies, which is used to learn representations of data with multilevel abstraction. The data has captured from the department of transportation; thus, the size of data is vast, and it can be analyzed to get the behavior of the traffic condition. The purpose of this paper is to monitor traffic behavior, which enables the user to make decisions to build the traffic-free cities. Experimental results show that the proposed methodology outperforms other existing methods such as KNN, CNN, and FIMT-DD. [ABSTRACT FROM AUTHOR]

The main purpose of the various methods of evaluating athlete feature recognition is to monitor the current health of the athletes, thereby providing some feedback on the quality of individual training. Based on deep learning and convolutional neural networks, this paper studies athlete target recognition and proposes a feature vector extraction method based on curvature zero point. Moreover, based on the ideas of deep learning and convolutional neural networks, this paper builds an athlete feature recognition model and optimizes the algorithm. In order to verify the feasibility and efficiency of feature extraction algorithm of the sport athletes proposed by this paper and to facilitate comparison with other algorithms, this paper conducts an algorithm performance test on the sport athlete database. The research results show that the method proposed in this paper has certain advantages in the feature extraction of athletes and can be used in subsequent sports training systems. [ABSTRACT FROM AUTHOR]

The transfer of scientific and technological achievements is an inevitable stage in the application of science and technology to the process of productivity. This process is accompanied by various influencing factors. How to eliminate the influence of adverse influence factors on the transformation of technology into productivity is crucial to the development of social productive forces. Based on this, from the perspective of deep learning, this study builds a technology transfer transformation platform through deep learning combined with data mining technology and analyzes the method in detail. On this basis, this paper takes a city as an example to analyze the platform of scientific and technological achievements transfer. In addition, by collecting existing data as system input and data mining analysis, this paper summarizes the advantages, disadvantages, opportunities and threats of the city's enterprises in the transformation of results and proposes corresponding countermeasures. The example verification shows that the method proposed in this study has certain practical effects and can provide theoretical reference for subsequent related research. [ABSTRACT FROM AUTHOR]

Identification/recognition of assault, fighting, shooting, and vandalism from video sequence using deep 2D and 3D convolutional neural networks (CNNs) is explored in this paper. Recent wave of extensive unrestricted urbanization has not only uplifted the standard of living, but has also threatened the safety of a common man leading to an extraordinary rise in crime rate. Although Closed-circuit television (CCTV) footage provides a monitoring framework, yet, it's useless without an auto volume crime detection system. The system proposed in this work is an effort to eradicate volume crimes through accurate detection in real-time. Firstly, a fine-grained annotated dataset including instance and activity information has been developed for real-world volume crimes. Secondly, a comparison between 3D CNN and 2D CNN network has been presented to identify the malicious event from the video sequence. This is carried out to explore the significance of spatial and temporal information present in the video for event recognition. It has been observed that 2D CNN even with lesser parameters achieved a promising classification accuracy of 91.2%and Area under the curve (AUC) of 95.2%on four classes. The system also reduces false alarm rate in comparison to state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

Email has sustained to be an essential part of our lives and as a means for better communication on the internet. The challenge pertains to the spam emails residing a large amount of space and bandwidth. The defect of state-of-the-art spam filtering methods like misclassification of genuine emails as spam (false positives) is the rising challenge to the internet world. Depending on the classification techniques, literature provides various algorithms for the classification of email spam. This paper tactics to develop a novel spam detection model for improved cybersecurity. The proposed model involves several phases like dataset acquisition, feature extraction, optimal feature selection, and detection. Initially, the benchmark dataset of email is collected that involves both text and image datasets. Next, the feature extraction is performed using two sets of features like text features and visual features. In the text features, Term Frequency-Inverse Document Frequency (TF-IDF) is extracted. For the visual features, color correlogram and Gray-Level Co-occurrence Matrix (GLCM) are determined. Since the length of the extracted feature vector seems to the long, the optimal feature selection process is done. The optimal feature selection is performed by a new meta-heuristic algorithm called Fitness Oriented Levy Improvement-based Dragonfly Algorithm (FLI-DA). Once the optimal features are selected, the detection is performed by the hybrid learning technique that is composed of two deep learning approaches named Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN). For improving the performance of existing deep learning approaches, the number of hidden neurons of RNN and CNN is optimized by the same FLI-DA. Finally, the optimized hybrid learning technique having CNN and RNN classifies the data into spam and ham. The experimental outcomes show the ability of the proposed method to perform the spam email classification based on improved deep learning. [ABSTRACT FROM AUTHOR]

The availability of techniques for driver distraction detection has been difficult to put to use because of delays caused due to lag in inferencing the model. Distractions caused due to handheld devices have been major causes of traffic accidents as they affect the decision-making capabilities of the driver and gives them less time to react to difficult situations. Often drivers try to multitask which reduces their reaction time leading to accidents, which can easily be avoided if they had been attentive. As such, problems related to the driver's negligence towards safety a possible solution is to monitor the driver and driving behavior and alerting them if they are distracted. In this paper, we propose a novel approach for detecting when a driver is distracted due to in hand electronic devices which is not only able to detect the distraction with high accuracy but also is energy and memory efficient. Our proposed compressed neural got an accuracy of 0.83 in comparison to 0.86 of heavyweight network. [ABSTRACT FROM AUTHOR]

Single biometric modalities like facial features and vein patterns despite being reliable characteristics show limitations that restrict them from offering high performance and robustness. Multimodal biometric systems have gained interest due to their ability to overcome the inherent limitations of the underlying single biometric modalities and generally have been shown to improve the overall performance for identification and recognition purposes. This paper proposes highly accurate and robust multimodal biometric identification as well as recognition systems based on fusion of face and finger vein modalities. The feature extraction for both face and finger vein is carried out by exploiting deep convolutional neural networks. The fusion process involves combining the extracted relevant features from the two modalities at score level. The experimental results over all considered public databases show a significant improvement in terms of identification and recognition accuracy as well as equal error rates. [ABSTRACT FROM AUTHOR]

In this paper, the authors propose an improved convolutional neural network for automatic arrhythmia classification using Electro-Cardio-Gram (ECG) signal. It is essential to periodically monitor the heart beat arrhythmia to reduce the risk of death due to cardiovascular disease (CVD). The Visual Geometry Group network (VGGNet) is being widely used in computer vision problems. However, the same network cannot be used for classification of ECG beats as ECG signal is different from image signal in terms of dimensionality and inherent features. Thus, the authors investigated the effect of decreasing depth and width of a convolutional neural network in context of cardiac arrhythmia classification. In this paper, six configurations differing in depth and width are evaluated using benchmark MIT-BIH database. A deep network having thirteen convolution layers but with a smaller number of filters (reduced width) showed outstanding performance for the given problem. Based on the findings, the work is further extended to propose an improved convolutional neural network named as modified VGGNet (mVGGNet) for the task of ECG arrhythmia classification into four classes which are normal (N), ventricular ectopic beat (VEB), supraventricular ectopic beat (SVEB) and fusion beat (F). The hyper-parameters of the proposed architecture were optimized using sequential model based global optimization (SMBO) algorithm. The proposed architecture is evaluated using subjected-oriented patient-independent evaluation protocol. The performance is evaluated using five-fold cross-validation. The proposed mVGGNet achieved 98.79% and 99.16% accuracy for ventricular ectopic beats (VEB) and supraventricular ectopic beats (SVEB) classification respectively. The proposed method resulted in higher specificity and precision as compared to other state-of-the-art algorithms. Thus, it can be effectively used for ECG arrhythmia classification. [ABSTRACT FROM AUTHOR]

BACKGROUND: The noise in magnetic resonance (MR) images causes severe issues for medical diagnosis purposes. OBJECTIVE: In this paper, we propose a discriminative learning based convolutional neural network denoiser to denoise the MR image data contaminated with noise. METHODS: The proposed method incorporates the use of depthwise separable convolution along with local response normalization with modified hyperparameters and internal skip connections to denoise the contaminated MR images. Moreover, the addition of parametric RELU instead of normal conventional RELU in our proposed architecture gives more stable and fine results. The denoised images were further segmented to test the appropriateness of the results. The network is trained on one dataset and tested on other dataset produces remarkably good results. RESULTS: Our proposed network was used to denoise the images of different noise levels, and it yields better performance as compared with various networks. The SSIM and PSNR showed an average improvement of (7.2 ± 0.002) % and (8.5 ± 0.25) % respectively when tested on different datasets without retaining the network. An improvement of 5% and 6% was achieved in the values of mean intersection over union (mIoU) and BF score when the denoised images were segmented for testing the relevancy in biomedical imaging applications. The statistical test suggests that the obtained results are statistically significant as p < 0.05. CONCLUSION: The denoised images obtained are more clinically suitable for medical image diagnosis purposes, as depicted by the evaluation parameters. Further, external clinical validation was performed by an experienced radiologist for testing the validation of the resulting images. [ABSTRACT FROM AUTHOR]

The encoding of a message is the creation of the message. The decoding of a message is how people can comprehend, and decipher the message. It is a procedure of understanding and interpretation of coded data into a comprehensible form. In this paper, a self-created explicitly defined function for encoding numerical digits into graphical representation is proposed. The proposed system integrates deep learning methods to get the probabilities of digit occurrence and Edge detection techniques for decoding the graphically encoded numerical digits to numerical digits as text. The proposed system's major objective is to take in an Image with digits encoded in graphical format and give the decoded stream of digits corresponding to the graph. This system also employs relevant pre-processing techniques to convert RGB to text and image to Canny image. Techniques such as Multi-Label Classification of images and Segmentation are used for getting the probability of occurrence. The dataset is created, on our own, that consists of 1000 images. The dataset has the training data and testing data in the proportion of 9 : 1. The proposed system was trained on 900 images and the testing was performed on 100 images which were ordered in 10 classes. The model has created a precision of 89% for probability prediction. [ABSTRACT FROM AUTHOR]

COVID-19 has been considered as a global pandemic. Recently, researchers are using deep learning networks for medical diseases' diagnosis. Some of these researches focuses on optimizing deep learning neural networks for enhancing the network accuracy. Optimizing the Convolutional Neural Network includes testing various networks which are obtained through manually configuring their hyperparameters, then the configuration with the highest accuracy is implemented. Each time a different database is used, a different combination of the hyperparameters is required. This paper introduces two COVID-19 diagnosing systems using both Residual Network and Xception Network optimized by random search in the purpose of finding optimal models that give better diagnosis rates for COVID-19. The proposed systems showed that hyperparameters tuning for the ResNet and the Xception Net using random search optimization give more accurate results than other techniques with accuracies 99.27536% and 100 % respectively. We can conclude that hyperparameters tuning using random search optimization for either the tuned Residual Network or the tuned Xception Network gives better accuracies than other techniques diagnosing COVID-19. [ABSTRACT FROM AUTHOR]

The squash function in capsule networks (CapsNets) dynamic routing is less capable of performing discrimination of non-informative capsules which leads to abnormal activation value distribution of capsules. In this paper, we propose vertical squash (VSquash) to improve the original squash by preventing the activation values of capsules in the primary capsule layer to shrink non-informative capsules, promote discriminative capsules and avoid high information sensitivity. Furthermore, a new neural network, (i) skip-connected convolutional capsule (S-CCCapsule), (ii) Integrated skip-connected convolutional capsules (ISCC) and (iii) Ensemble skip-connected convolutional capsules (ESCC) based on CapsNets are presented where the VSquash is applied in the dynamic routing. In order to achieve uniform distribution of coupling coefficient of probabilities between capsules, we use the Sigmoid function rather than Softmax function. Experiments on Guangzhou Women and Children's Medical Center (GWCMC), Radiological Society of North America (RSNA) and Mendeley CXR Pneumonia datasets were performed to validate the effectiveness of our proposed methods. We found that our proposed methods produce better accuracy compared to other methods based on model evaluation metrics such as confusion matrix, sensitivity, specificity and Area under the curve (AUC). Our method for pneumonia detection performs better than practicing radiologists. It minimizes human error and reduces diagnosis time. [ABSTRACT FROM AUTHOR]

In this paper, a comprehensive Convolutional Neural Network (CNN) based classifier "WAF-LeNet" is proposed and developed to be used in traffic signs recognition and identification as an empowerment of autonomous driving technologies. The implemented architecture is a deep fifteen-layer network that has been selected after extensive trials to be fast enough to suit the designated application. The CNN got trained using Adam's optimization algorithm as a variant of the Stochastic Gradient Descent (SGD) technique. The learning process is carried out using the well-known "German Traffic Sign Dataset – GTSRB". The data has been partitioned into training, validation and testing data sets. Additionally, more random traffic signs images are collected from the web and further used to test the robustness of the proposed CNN classifier. The paper goes through the development process in details and shows the image processing pipeline harnessed in the development. The proposed approach proved successful in identifying correctly 96.5% of the testing data set and 100% of the robustness data set with much smaller and faster network than other counterparts. [ABSTRACT FROM AUTHOR]

The identification of landmark features such as optic disc is of high prognostic significance in diagnosing various ophthalmic diseases. A retinal fundus photograph provides a non-invasive observation of the optic disc. The wide variability present in fundus images poses difficulties in its detection and further analysis. The reported work is a part of the fundus image screening for the diagnosis of Retinopathy of Prematurity (ROP), a sight threatening disorder seen in preterm infants. The diagnostic procedure for this disease estimates blood vessel tortuosity in a pre-defined area around the optic disc. Hence accurate optic disc localization is very important for the disease diagnosis. In this paper, we present an optic disc localization technique using a deep neural network based framework. The proposed system relies on the underlying architecture of YOLOv3, a fully convolutional neural network pipeline for object detection and localization. The new approach is tested in 10 different data sets and has achieved an overall accuracy of 99.25%, outperforming other deep learning-based OD detection methods. The test results guarantees the robustness of the proposed technique, and hence may be deployed to assist medical experts for disease diagnosis. [ABSTRACT FROM AUTHOR]

In order to meet the real-time and robustness requirement of driverless cars driving on highway, this paper proposed a lane line identification method based on the Deep Learning. This method first built a lane line image library and then input the pictures of the lane line image for denoising and normalized processing. Secondly, the Lenet – 5 network model was used for classification and recognition, with a recognition rate of 99.4%, and the lane line type was displayed through GUI interface. Finally, this method was compared with the support vector machine and BP neural network, and the results effectively verified that the method can satisfy the requirement of real-time and accuracy of lane line identification. [ABSTRACT FROM AUTHOR]

In this paper, we extend the discriminant correlation filter (DCF) based deep learning tracker to multi-object tracking. For each object, we use an individual tracker to estimate the position. Two different pre-trained networks are used as feature extractors, respectively. The response peak and oscillation are both considered to validate the tracking. When the object is lost, the discriminative appearance model achieved by DCF is considered as a part of the feature representation between the object and detection for data association. In order to validate our method, we analyze and test our approach on the MOT2D2015 and MOT17 benchmarks for multiple pedestrian tracking. The results show that our approach performs superiorly against several recent state-of-the-art online multi-object trackers. [ABSTRACT FROM AUTHOR]

Deep learning is a field of Artificial Intelligence that has recently drawn a lot of attention with the desire to build up a quick, automatic and accurate system for image identification and classification. Deep learning serves as a fundamental part of modern computer vision solutions. However, as the architectures become deep and powerful new challenges in the process of training emerge. This includes the computational cost associated with training deep and large networks. In this work, the focus is on pruning and evaluation of state-of-the-art deep convolutional neural network for image-based plant disease and plants species classification. Pruning filters allow the reduction of parameters by removing unimportant filters and its feature maps. In this paper, the performance of pruned networks is evaluated across three datasets. It is observed that pruned DenseNet with Self-Normalization Neural Network (SNN) approach learns 2x faster compared to the initial DenseNet architecture. Additionally, pruning filters allow the reduction of the number of parameters and FLOPs by approximately 14% and 25% respectively. The aim is to create a fast and efficient model for the purpose of identification of plant diseases. Fast methods are desired for early identifications of diseases before damages occur. The proposed method achieves a satisfactory accuracy performance on PlantVillage, LeafSnap and Swedish-leaf dataset using held-out dataset. Our best pruned model gives an accuracy of 99.24%, 86.64%, and 97.5% on PlantVillage, LeafSnap, and Swedish-leaf datasets respectively. [ABSTRACT FROM AUTHOR]

This paper presents a fully-automated method to detect retinal abnormalities in both a global and local context by generating: (1) a disease label with a probability score and (2) a 4-channel pixel-level segmentation map of retinal lesions. The characteristics of retinal abnormalities, which occur as various shapes, sizes, and distribution at different regions, are a challenge in accomplishing these tasks. In addition, the small amount of image-level labelled images in public databases and the unavailability of lesion-level annotations for most of these publicly available images also pose as challenges. These shortcomings motivate our exploration of various CNN architectures to extract multi-scale contextual information, such that we investigate the impact of different arrangements of multi-sized convolutional kernels appended to a modified pre-trained encoder. Additionally, to prevent the loss of detailed information for small lesions, we exploit the advantages of feature map concatenation from the output of these multi-scale convolutions to its corresponding decoder layer. A new two-phase training strategy is also implemented to tackle the problem of dataset imbalance between image-level label and lesion-level label classes. The direct comparison between our proposed methods and currently published state-of-the-art methods with the same databases confirms that our best model outperforms existing published methods. [ABSTRACT FROM AUTHOR]

Image super resolution has gained a lot of attention due to its applications in different fields of image processing. It is used to produce high-resolution images from low-resolution input. Because of the excellent learning capability of convolution neural networks, these networks are able to learn complex spatial structures for image super-resolution. In this paper, two different architectures have been proposed for image super resolution. The first architecture is Dual Subpixel Layer Convolution Neural Network (DSL-CNN), which stacks two subpixel CNN architectures to enhance model depth for better representational capability. Two stages provide an effective upscaling factor of 4. In the second architecture, named as Residue based Dual Subpixel Layer Convolution Neural Network (RDSL-CNN), two-stage residual learning has been introduced which effectively sustains the high frequency details and provides superior results than the previous state-of-the-art methods. The performance of the two architectures has been evaluated on various image datasets, and compared with other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Plant species recognition from images or videos is challenging due to a large diversity of plants, variation in orientation, viewpoint, background clutter, etc. In this paper, plant species recognition is carried out using two approaches, namely, traditional method and deep learning approach. In traditional method, feature extraction is carried out using Hu moments (shape features), Haralick texture, local binary pattern (LBP) (texture features) and color channel statistics (color features). The extracted features are classified using different classifiers (linear discriminant analysis, logistic regression, classification and regression tree, naïve Bayes, k-nearest neighbor, random forest and bagging classifier). Also, different deep learning architectures are tested in the context of plant species recognition. Three standard datasets (Folio, Swedish leaf and Flavia) and one real-time dataset (Leaf12) is used. It is observed that, in traditional method, feature vector obtained by the combination of color channel statistics+LBP+Hu+Haralick with Random Forest classifier for Leaf12 dataset resulted in a plant recognition accuracy (rank-1) of 82.38%. VGG 16 Convolutional Neural Network (CNN) architecture with logistic regression resulted in an accuracy of 97.14% for Leaf12 dataset. An accuracy of 96.53%, 96.25% and 99.41% is obtained for Folio, Flavia and Swedish leaf datasets using VGG 19 CNN architecture with logistic regression as a classifier. It is also observed that the VGG (Very large Convolutional Neural Network) CNN models provided a higher accuracy rate compared to traditional methods. [ABSTRACT FROM AUTHOR]

Deep learning algorithms have recently been applied to solving challenging problems in medicine such as medical image classification and analysis. In some areas, those algorithms have outperformed the human medical experts experience in diagnosis. Thus, in this paper we apply three different deep networks to solve the problem of brain hemorrhage identification in CT images. The motivation behind this work is the difficulty that radiologists encounter when diagnosing a hemorrhagic brain CT image, in particularly in the early stages of the brain bleeding. Autoencoder (AE), stacked autoencoder (SAE), and convolutional neural network (CNN) are employed and trained to classify the CT images into hemorrhagic or non-hemorrhagic. Experimentally, it was found that all employed networks performed differently in terms of accuracy, error reached, and training time. However, stacked autoencoder has achieved a higher accuracy and lesser error compared to other used networks. [ABSTRACT FROM AUTHOR]

The online smartphone-based human activity recognition (HAR) has a variety of applications such as fitness tracking, healthcare…etc. Currently, the signals generated from smartphone-embedded sensors are used for HAR systems. The smartphone-embedded sensors are utilized in order to provide an unobtrusive platform for HAR. In this paper, we propose a deep convolution neural network (CNN) model that provides an effective and efficient smartphone-based HAR system. For automatic local features extraction from the raw time-series data, we use the CNN while simple time-domain statistical features are used to extract more distinguishable features. Furthermore, we explore the impact of a novel data augmentation on the recognition accuracy of the proposed model. The performance of the proposed method is evaluated using two public data sets (UCI and WISDM) which are collected using smartphones. Experimentally, we show how the proposed model establishes the state-of-the-art performance using these datasets. Finally, to demonstrate the applicability of the proposed model for online smartphone-based HAR, the computational cost of the model is evaluated. [ABSTRACT FROM AUTHOR]

Task-specific word identification aims to choose the task-related words that best describe a short text. Existing approaches require well-defined seed words or lexical dictionaries (e.g., WordNet), which are often unavailable for many applications such as social discrimination detection and fake review detection. However, we often have a set of labeled short texts where each short text has a task-related class label, e.g., discriminatory or non-discriminatory, specified by users or learned by classification algorithms. In this paper, we focus on identifying task-specific words and phrases from short texts by exploiting their class labels rather than using seed words or lexical dictionaries. We consider the task-specific word and phrase identification as feature learning. We train a convolutional neural network over a set of labeled texts and use score vectors to localize the task-specific words and phrases. Experimental results on sentiment word identification show that our approach significantly outperforms existing methods. We further conduct two case studies to show the effectiveness of our approach. One case study on a crawled tweets dataset demonstrates that our approach can successfully capture the discrimination-related words/phrases. The other case study on fake review detection shows that our approach can identify the fake-review words/phrases. [ABSTRACT FROM AUTHOR]