Feature Extraction Stage Research Articles

An AI-Inspired Spatio-Temporal Neural Network for EEG-Based Emotional Status.

The accurate identification of the human emotional status is crucial for an efficient human-robot interaction (HRI). As such, we have witnessed extensive research efforts made in developing robust and accurate brain-computer interfacing models based on diverse biosignals. In particular, previous research has shown that an Electroencephalogram (EEG) can provide deep insight into the state of emotion. Recently, various handcrafted and deep neural network (DNN) models were proposed by researchers for extracting emotion-relevant features, which offer limited robustness to noise that leads to reduced precision and increased computational complexity. The DNN models developed to date were shown to be efficient in extracting robust features relevant to emotion classification; however, their massive feature dimensionality problem leads to a high computational load. In this paper, we propose a bag-of-hybrid-deep-features (BoHDF) extraction model for classifying EEG signals into their respective emotion class. The invariance and robustness of the BoHDF is further enhanced by transforming EEG signals into 2D spectrograms before the feature extraction stage. Such a time-frequency representation fits well with the time-varying behavior of EEG patterns. Here, we propose to combine the deep features from the GoogLeNet fully connected layer (one of the simplest DNN models) together with the OMTLBP_SMC texture-based features, which we recently developed, followed by a K-nearest neighbor (KNN) clustering algorithm. The proposed model, when evaluated on the DEAP and SEED databases, achieves a 93.83 and 96.95% recognition accuracy, respectively. The experimental results using the proposed BoHDF-based algorithm show an improved performance in comparison to previously reported works with similar setups.

Combining convolutional neural networks and self-attention for fundus diseases identification

Early detection of lesions is of great significance for treating fundus diseases. Fundus photography is an effective and convenient screening technique by which common fundus diseases can be detected. In this study, we use color fundus images to distinguish among multiple fundus diseases. Existing research on fundus disease classification has achieved some success through deep learning techniques, but there is still much room for improvement in model evaluation metrics using only deep convolutional neural network (CNN) architectures with limited global modeling ability; the simultaneous diagnosis of multiple fundus diseases still faces great challenges. Therefore, given that the self-attention (SA) model with a global receptive field may have robust global-level feature modeling ability, we propose a multistage fundus image classification model MBSaNet which combines CNN and SA mechanism. The convolution block extracts the local information of the fundus image, and the SA module further captures the complex relationships between different spatial positions, thereby directly detecting one or more fundus diseases in retinal fundus image. In the initial stage of feature extraction, we propose a multiscale feature fusion stem, which uses convolutional kernels of different scales to extract low-level features of the input image and fuse them to improve recognition accuracy. The training and testing were performed based on the ODIR-5k dataset. The experimental results show that MBSaNet achieves state-of-the-art performance with fewer parameters. The wide range of diseases and different fundus image collection conditions confirmed the applicability of MBSaNet.

Bayesian Deep Learning for Hyperspectral Image Classification With Low Uncertainty

In recent years, deep learning models have been widely used for hyperspectral image (HSI) classification and most of existing deep learning-based methods merely focused on high classification accuracy. However, in real applications, classification with low uncertainty matters as much as accurate classification. Unfortunately, existing methods fail to consider uncertainty. To tackle this challenge, for the first time, Bayesian deep learning (BDL) is investigated to analyze the model uncertainty for HSI classification. Specifically, first, at the feature extraction stage, an HSI classification framework based on BDL, which contains two Bayesian Gabor layers and a global pooling layer (i.e., BDL-G <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), is proposed. In BDL-G <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , parameters in Gabor layers are sampled from the Gaussian distribution. The proposed BDL-G <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> not only provides the uncertainty estimation, but also strengthens the structure characteristic (i.e., texture) of HSI. Second, to model the uncertainty at the final classification stage, BDL-G <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> is combined with a Bayesian fully-connected layer (i.e., BDL-G <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> -BFL), where the parameters’ distribution is adjusted adaptively. In the proposed BDL-G <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> -BFL, the uncertainty at feature extraction and classification stages are both captured, and a whole uncertainty estimation framework is established. Experimental results on the three public HSI datasets demonstrates the superiority in both accuracy and uncertainty. The proposed Bayesian deep learning-based methods pioneer a new direction and provide useful inspiration and experience for practical applications.

Learning Cross-Channel Representations for Semantic Segmentation

Semantic segmentation is a fundamental problem in multimedia which requires delicate per-pixel predictions of object categories. Recently, many researchers strive to refine the pixel-wise feature with spatial-contextual information. However, many of them still neglect the invisible hand of cross-channel information which provides inherent semantics to facilitate the segmentation performance. On the one hand, in the feature extraction stage, enhancing informative channels and suppressing trivial ones contribute to the acquisition of valuable semantic features, and thus improving the segmentation accuracy. On the other hand, in the prediction stage, we can predict the complete objects more clearly by finding the connections and complements between different channels, which can also contribute to the pixel prediction. And based on this idea, we propose a novel Channel-Adaptive Network for semantic segmentation, which is capable of enhancing the features from the perspective of channels in both feature extraction stage and prediction stage. Specifically, we propose two modules: (i) the Comprehensive Information Channel Attention (CiCA) module that addresses the shortcomings of existing channel attention by learning both low and high frequency components within each channel for emphasizing the informative channels; (ii) the Inter-Channel Relationship Reasoning (iCRR) module which is applied on the top of the feature extractor to adaptively enhance the interdependent channels by mining the complementary associations between them. Besides, our Channel-Adaptive Network is highly flexible, with a plug-and-play design. Extensive experiments have demonstrated that our method achieves the state-of-the-art segmentation performance on three challenging datasets, including Cityscapes (82.1%), ADE20K (46.51%) and PASCAL Context (55.0%).

A visual transformer-based smart textual extraction method for financial invoices.

In era of big data, the computer vision-assisted textual extraction techniques for financial invoices have been a major concern. Currently, such tasks are mainly implemented via traditional image processing techniques. However, they highly rely on manual feature extraction and are mainly developed for specific financial invoice scenes. The general applicability and robustness are the major challenges faced by them. As consequence, deep learning can adaptively learn feature representation for different scenes and be utilized to deal with the above issue. As a consequence, this work introduces a classic pre-training model named visual transformer to construct a lightweight recognition model for this purpose. First, we use image processing technology to preprocess the bill image. Then, we use a sequence transduction model to extract information. The sequence transduction model uses a visual transformer structure. In the stage target location, the horizontal-vertical projection method is used to segment the individual characters, and the template matching is used to normalize the characters. In the stage of feature extraction, the transformer structure is adopted to capture relationship among fine-grained features through multi-head attention mechanism. On this basis, a text classification procedure is designed to output detection results. Finally, experiments on a real-world dataset are carried out to evaluate performance of the proposal and the obtained results well show the superiority of it. Experimental results show that this method has high accuracy and robustness in extracting financial bill information.

A Hybrid Method of Feature Extraction for Signatures Verification Using CNN and HOG a Multi-Classification Approach

The offline signature verification system’s feature extraction stage is regarded as crucial and has a big impact on how well these systems perform because the quantity and calibre of the features that are extracted determine how well these systems can distinguish between authentic and fake signatures. In this study, we introduced a hybrid method for extracting features from signature images, wherein Convolutional Neural Network (CNN) and Histogram of Oriented Gradients (HOG) were used, followed by the feature selection algorithm (Decision Trees) to identify the key features, and finally the CNN and HOG methods were combined. Three classifiers were employed to evaluate the hybrid method’s efficacy (long short-term memory, support vector machine, and K-nearest Neighbor). The experimental findings indicated that our suggested model executed quite satisfactorily in terms of efficiency and predictive ability, obtaining accuracy of (95.4%, 95.2%, and 92.7%) with UTSig dataset, and (93.7%, 94.1%, and 91.3%) with CEDAR dataset. This accuracy is deemed to be a high significance, particularly given that we checked skilled forged signatures that are more difficult to recognize than other forms of forged signatures like (simple or opposite-hand).

Meta-Heuristic Optimized Hybrid Wavelet Features for Arrhythmia Classification

The non-invasive evaluation of the heart through EectroCardioGraphy (ECG) has played a key role in detecting heart disease. The analysis of ECG signals requires years of learning and experience to interpret and extract useful information from them. Thus, a computerized system is needed to classify ECG signals with more accurate results effectively. Abnormal heart rhythms are called arrhythmias and cause sudden cardiac deaths. In this work, a Computerized Abnormal Heart Rhythms Detection (CAHRD) system is developed using ECG signals. It consists of four stages; preprocessing, feature extraction, feature optimization and classifier. At first, Pan and Tompkins algorithm is employed to detect the envelope of Q, R and S waves in the preprocessing stage. It uses a recursive filter to eliminate muscle noise, <i>T</i>-wave interference and baseline wander. As the analysis of ECG signal in the spatial domain does not provide a complete description of the signal, the feature extraction involves using frequency contents obtained from multiple wavelet filters; bi-orthogonal, Symlet and Daubechies at different resolution levels in the feature extraction stage. Then, Black Widow Optimization (BWO) is applied to optimize the hybrid wavelet features in the feature optimization stage. Finally, a kernel based Support Vector Machine (SVM) is employed to classify heartbeats into five classes. In SVM, Radial Basis Function (RBF), polynomial and linear kernels are used. A total of ∼15000 ECG signals are obtained from the Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database for performance evaluation of the proposed CAHRD system. Results show that the proposed CAHRD system proved to be a powerful tool for ECG analysis. It correctly classifies five classes of heartbeats with 99.91% accuracy using an RBF kernel with 2^nd level wavelet coefficients. The CAHRD system achieves an improvement of ∼6% over random projections with the ensemble SVM approach and ∼2% over morphological and ECG segment based features with the RBF classifier.

An Automatic Deep Neural Network Model for Fingerprint Classification

The accuracy of fingerprint recognition model is extremely important due to its usage in forensic and security fields. Any fingerprint recognition system has particular network architecture whereas many other networks achieve higher accuracy. To solve this problem in a unified model, this paper proposes a model that can automatically specify itself. So, it is called an automatic deep neural network (ADNN). Our algorithm can specify the appropriate architecture of the neural network used and some significant parameters of this network. These parameters are the number of filters, epochs, and iterations. It guarantees the highest accuracy by updating itself until achieving 99% accuracy then it stops and outputs the result. Moreover, this paper proposes an end-to-end methodology for recognizing a person’s identity from the input fingerprint image based on a residual convolutional neural network. It is a complete system and is fully automated whether in the features extraction stage or the classification stage. Our goal is to automate this fingerprint recognition system because the more automatic the system is, the more time and effort it saves. Our model also allows users to react by inputting the initial values of these parameters. Then, the model updates itself until it finds the optimal values for the parameters and achieves the best accuracy. Another advantage of our algorithm is that it can recognize people from their thumb and other fingers and its ability to recognize distorted samples. Our algorithm achieved 99.75% accuracy on the public fingerprint dataset (SOCOFing). This is the best accuracy compared with other models.

COMPREHENSIVE STUDY OF DEEP LEARNING BASED TELUGU OCR

The aim of the project is to understand offline One of the most popular and difficult pattern recognition subjects is the use of optical character recognition (OCR) to read handwritten Telugu letters. This study suggests a three-stage OCR solution for Telugu documents that includes pre-processing, feature extraction, and classification. For the extraction of boundary edge pixel points during preprocessing, we used median filtering on the input characters as well as normalisation and skeletonization techniques. Each character is initially divided into three 3x3 grids during the feature extraction stage, and the associated centroid for each of the nine zones is assessed. This allows us to recognise characters in various styles. Following that, we drew the projection angel's horizontal and vertical symmetry to the character's closest pixel.

Novel Framework of Segmentation 3D MRI of Brain Tumors

Medical image segmentation is a crucial process for computer-aided diagnosis and surgery. Medical image segmentation refers to portioning the images into small, disjointed parts for simplifying the processes of analysis and examination. Rician and speckle noise are different types of noise in magnetic resonance imaging (MRI) that affect the accuracy of the segmentation process negatively. Therefore, image enhancement has a significant role in MRI segmentation. This paper proposes a novel framework that uses 3D MRI images from Kaggle and applies different diverse models to remove Rician and speckle noise using the best possible noise-free image. The proposed techniques consider the values of Peak Signal to Noise Ratio (PSNR) and the level of noise as inputs to the attention-U-Net model for segmentation of the tumor. The framework has been divided into three stages: removing speckle and Rician noise, the segmentation stage, and the feature extraction stage. The framework presents solutions for each problem at a different stage of the segmentation. In the first stage, the framework uses Vibrational Mode Decomposition (VMD) along with Block-matching and 3D filtering (Bm3D) algorithms to remove the Rician. Afterwards, the most significant Rician noise-free images are passed to the three different methods: Deep Residual Network (DeRNet), Dilated Convolution Auto-encoder Denoising Network (Di-Conv-AE-Net), and Denoising Generative Adversarial Network (DGAN-Net) for removing the speckle noise. VMD and Bm3D have achieved PSNR values for levels of noise (0, 0.25, 0.5, 0.75) for reducing the Rician noise by (35.243, 32.135, 28.214, 24.124) and (36.11, 31.212, 26.215, 24.123) respectively. The framework also achieved PSNR values for removing the speckle noise process for each level as follows: (34.146, 30.313, 28.125, 24.001), (33.112, 29.103, 27.110, 24.194), and (32.113, 28.017, 26.193, 23.121) for DeRNet, Di-Conv-AE-Net, and DGAN-Net, respectively. The experiments that have been conducted have proved the efficiency of the proposed framework against classical filters such as Bilateral, Frost, Kuan, and Lee according to different levels of noise. The attention gate U-Net achieved 94.66 and 95.03 in the segmentation of free noise images in dice and accuracy, respectively.

Tropical Cyclogenesis Detection From Remotely Sensed Sea Surface Winds Using Graphical and Statistical Features-Based Broad Learning System

This paper proposed a graphical and statistical features-based broad learning system (GSF-BLS) to detect tropical cyclogenesis with the Cross-Calibrated Multi-Platform version 2.0 (CCMP V2.0) wind products. The framework of the proposed model is composed of three modules: the data preprocessing module, the feature extraction module, and the basis broad learning system (BLS). At the stage of data preprocessing, we use the CCMP V2.0 data to match the best tracks and the Global Tropical Cloud Cluster tracks to obtain the developed and undeveloped samples. At the feature extraction stage, a convolution module with pre-trained weights is used to extract the graphical features (GFs). Meanwhile, the statistical features (SFs) are calculated based on the divided sub-regions of each sample. Thus, the combination of these GFs and SFs forms the input vectors. Then, the training time of GSF-BLS on CPU is only one-twentieth of that of deep learning models, showing its simplicity and efficiency in model training. The overall accuracy, probability of detection (POD), and false alarm rate (FAR) on the testing set are 89.46%, 86.78%, and 8.31%, respectively. More importantly, the incremental learning ability of GSF-BLS makes it superior to most deep learning models in model updating, which can avoid the computational burden caused by retraining. Finally, the case study results show that GSF-BLS can predict tropical cyclogenesis in 52 of 70 cases in advance, and the average lead times are 13.54 hours. Therefore, the experimental results demonstrate that GSF-BLS is a promising tropical cyclogenesis detection model.

Capturing Small Objects and Edges Information for Cross-Sensor and Cross-Region Land Cover Semantic Segmentation in Arid Areas

In the oasis area adjacent to the desert, there is more complex land cover information with rich details, multiscales of interest objects, and blur edge information, which poses some challenges to the semantic segmentation task in remote sensing images (RSIs). In traditional semantic segmentation methods, detailed spatial information is more likely lost in feature extraction stage and the global context information is more effectively integrated into segmentation results. To overcome these land cover semantic segmentation model, FPN_PSA_DLV3+ network, is proposed in an encoder–decoder manner capturing more fine edge and small objects information in RSIs. In the encoder stage, the improved atrous spatial pyramid pooling module extracts the multiscale features, especially small-scale feature details; feature pyramid network (FPN) module realizes better integration of detailed information and semantic information; and the spatial context information at both global and local levels is enhanced by introducing polarized self-attention (PSA) module. For the decoder stage, the FPN_PSA_DLV3+ network further adds a feature fusion branch to concatenate more low-level features. We select Landsat5/7/8 satellite RSIs from the areas of north and south of Xinjiang. Then, three self-annotated time-series datasets with more small objects and fine edges information are constructed by data augmentation. The experimental results show that the proposed method improves the segmentation performance of small targets and edges, and the classification performance increases from 81.55% to 83.10% <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">F</i> 1 score and from 72.65% to 74.82% mean intersection over union only using red–green–blue bands. Meanwhile, the FPN_PSA_DLV3+ network shows great generalization in cross region and cross sensor.

Advertising Image Saliency Prediction Method Based on Score Level Fusion

At present, visual saliency prediction algorithms have been developed more and more mature, but most of the current saliency prediction algorithms are aimed at natural images. Due to the inconsistency of elements and features between natural images and advertising images, the existing saliency prediction algorithms show poor robustness and low inference speed to advertising images, which severely limits its commercial application in advertising design and evaluation. In view of this, a saliency prediction algorithm for advertisement images is proposed in this paper. In the feature extraction stage, two text candidate regions based on intensity feature and improved MESR algorithm are first obtained and further integrated to produce a two-dimensional text confidence score. Meanwhile, a saliency confidence score is also obtained by an improved natural image saliency prediction network. Then, the score level fusion strategy was adopted to fuse the two confidence scores to get the final saliency prediction map. The experimental results show that the proposed model has good accuracy and robustness in advertising images, as well as the most remarkable inference speed, which can meet the demand for real-time performance of advertising image saliency prediction, leading to great practical and commercial value.

Model-Aware Pre-Training for Radial Distortion Rectification.

Camera lenses often suffer from optical aberrations, causing radial distortion in the captured images. In those images, there exists a clear and general physical distortion model. However, in existing solutions, such rich geometric prior is under-utilized, and the formulation of an effective prediction target is under-explored. To this end, we introduce Radial Distortion TRansformer (RDTR), a new framework for radial distortion rectification. Our RDTR includes a model-aware pre-training stage for distortion feature extraction and a deformation estimation stage for distortion rectification. Technically, on the one hand, we formulate the general radial distortion (i.e., barrel distortion and pincushion distortion) in camera-captured images with a shared geometric distortion model and perform a unified model-aware pre-training for its learning. With the pre-training, the network is capable of encoding the specific distortion pattern of a radially distorted image. After that, we transfer the learned representations to the learning of distortion rectification. On the other hand, we introduce a new prediction target called backward warping flow for rectifying images with any resolution while avoiding image defects. Extensive experiments are conducted on our synthetic dataset, and the results demonstrate that our method achieves state-of-the-art performance while operating in real-time. Besides, we also validate the generalization of RDTR on real-world images. Our source code and the proposed dataset are publicly available at https://github.com/wwd-ustc/RDTR.

Computational Linguistics with Optimal Deep Belief Network Based Irony Detection in Social Media

Computational linguistics refers to an interdisciplinary field associated with the computational modelling of natural language and studying appropriate computational methods for linguistic questions. The number of social media users has been increasing over the last few years, which have allured researchers’ interest in scrutinizing the new kind of creative language utilized on the Internet to explore communication and human opinions in a better way. Irony and sarcasm detection is a complex task in Natural Language Processing (NLP). Irony detection has inferences in advertising, sentiment analysis (SA), and opinion mining. For the last few years, irony-aware SA has gained significant computational treatment owing to the prevalence of irony in web content. Therefore, this study develops Computational Linguistics with Optimal Deep Belief Network based Irony Detection and Classification (CLODBN-IRC) model on social media. The presented CLODBN-IRC model mainly focuses on the identification and classification of irony that exists in social media. To attain this, the presented CLODBN-IRC model performs different stages of pre-processing and TF-IDF feature extraction. For irony detection and classification, the DBN model is exploited in this work. At last, the hyperparameters of the DBN model are optimally modified by improved artificial bee colony optimization (IABC) algorithm. The experimental validation of the presented CLODBN-IRC method can be tested by making use of benchmark dataset. The simulation outcomes highlight the superior outcomes of the presented CLODBN-IRC model over other approaches.

Ensemble of Local Texture Descriptor for Accurate Breast Cancer Detection from Histopathologic Images

Histopathological analysis is important for detection of the breast cancer (BC). Computer-aided diagnosis and detection systems are developed to assist the radiologist in the diagnosis process and to relieve the patient from unnecessary pain. In this study, a computer-aided diagnosis system for the early detection of benign and malignant breast cancer is proposed. The proposed system consists of feature extraction, feature ensemble, and classification stages. Various preprocessing steps such as grayscale conversion, noise filtering, and image resizing are employed on the input histopathological images. The local texture descriptors namely Local Binary Pattern (LBP), Frequency Decoded LBP (FDLBP), Binary Gabor Pattern (BGP), Local Phase Quantization (LPQ), Binarized Statistical Image Features (BSIF), CENsus TRansform hISTogram (CENTRIST), and Pyramid Histogram of Oriented Gradients (PHOG) are employed for feature extraction from the histopathologic images. The obtained features are then concatenated for the construction of the ensemble of the features. Three classifiers namely Support Vector Machines (SVM), K-nearest neighbor (KNN), and Neural Networks (NN) are used in the detection of the BC and the classification accuracy score is used for performance evaluation. A dataset called BreaKHis is used in the studies. There are 9109 microscopic pictures in BreaKHis, with 2480 benign samples and 5429 malignant samples. During the collection of the data, 82 patients' breast tumor tissues were envisioned using various magnification factors such as 40X, 100X, 200X, and 400X. The accuracy score is used to assess the acquired findings. The results show that the proposed method has the potential to use accurate BC detection.

Space Object Recognition With Stacking of CoAtNets Using Fusion of RGB and Depth Images

Space situational awareness (SSA) system requires recognition of space objects that are varied in sizes, shapes, and types. The space images are challenging because of several factors such as illumination and noise and thus make the recognition task complex. Image fusion is an important area in image processing for various applications including RGB-D sensor fusion, remote sensing, medical diagnostics, and infrared and visible image fusion. Recently, various image fusion algorithms have been developed and they showed a superior performance to explore more information that are not available in single images. In this paper, we compared various methods of RGB and Depth image fusion for space object classification task. The experiments were carried out, and the performance was evaluated using 13 fusion performance metrics. It was found that the guided filter context enhancement (GFCE) outperformed other image fusion methods in terms of average gradient (8.2593), spatial frequency (28.4114), and entropy (6.9486). additionally, due to its ability to balance between good performance and inference speed (11.41 second), GFCE was selected for RGB and Depth image fusion stage before feature extraction and classification stage. The outcome of fusion method is fused images that were used to train a deep ensemble of CoAtNets to classify space objects into ten categories. The deep ensemble learning methods including bagging, boosting, and stacking were trained and evaluated for classification purposes. It was found that combination of fusion and stacking was able to improve classification accuracy largely compared to the baseline methods by producing an average accuracy of 89 % and average F1 score of 89 %.

Performance Analysis of Breast Cancer Detection Method Using ANFIS Classification Approach

Breast cancer is one of the deadly diseases prevailing in women. Earlier detection and diagnosis might prevent the death rate. Effective diagnosis of breast cancer remains a significant challenge, and early diagnosis is essential to avoid the most severe manifestations of the disease. The existing systems have computational complexity and classification accuracy problems over various breast cancer databases. In order to overcome the above-mentioned issues, this work introduces an efficient classification and segmentation process. Hence, there is a requirement for developing a fully automatic methodology for screening the cancer regions. This paper develops a fully automated method for breast cancer detection and segmentation utilizing Adaptive Neuro Fuzzy Inference System (ANFIS) classification technique. This proposed technique comprises preprocessing, feature extraction, classifications, and segmentation stages. Here, the wavelet-based enhancement method has been employed as the preprocessing method. The texture and statistical features have been extracted from the enhanced image. Then, the ANFIS classification algorithm is used to classify the mammogram image into normal, benign, and malignant cases. Then, morphological processing is performed on malignant mammogram images to segment cancer regions. Performance analysis and comparisons are made with conventional methods. The experimental result proves that the proposed ANFIS algorithm provides better classification performance in terms of higher accuracy than the existing algorithms.

A Model for Qur’anic Sign Language Recognition Based on Deep Learning Algorithms

Deaf and dumb Muslims cannot reach advanced levels of education due to the impact of obstruction on their educational attainment. This leads to their inability to learn, recite, and understand the meanings and interpretations of the Holy Qur’an as easily as ordinary people, which also prevents them from applying Islamic rituals such as prayer that require learning and reading the Holy Qur’an. In this paper, we propose a new model for Qur’anic sign language recognition based on convolutional neural networks through data preparation, preprocessing, feature extraction, and classification stages. The proposed model is aimed at recognizing the movements of the Arabic sign language by recognizing the hand gestures that refer to the dashed Qur’anic letters in order to help the deaf and dumb learn their Islamic rituals. The experiments have been conducted on a part of a large Arabic sign language dataset called ArSL2018, which represents the 14 dashed letters in the Holy Qur’an, so that this part contains only 24,137 images. The experimental results demonstrate that the proposed model performs better than the other existing models.

Spatial Pyramid Pooling and Adaptively Feature Fusion based Yolov3 for Traffic Sign Detection

Traffic sign detection is a key part of intelligent assisted driving, but also a challenging task due to the small size and different scales of objects in foreground and closed range. In this paper, we propose a new traffic sign detection scheme: Spatial Pyramid Pooling and Adaptively Spatial Feature Fusion based Yolov3 (SPP and ASFF-Yolov3). In order to integrate the target detail features and environment context features in the feature extraction stage of Yolov3 network, the Spatial Pyramid Pooling module is introduced into the pyramid network of Yolov3. Additionally, Adaptively Spatial Feature Fusion module is added to the target detection phase of the pyramid network of Yolov3 to avoid the interference of different scale features with the process of gradient calculation. Experimental results show the effectiveness of the proposed SPP and ASFF-Yolov3 network, which achieves better detection results than the original Yolov3 network. It can archive real-time inference speed despite inferior to the original Yolov3 network. The proposed scheme will add an option to the solutions of traffic sign detection with real-time inference speed and effective detection results.