In this research, plant pathogens are considered as big data because of the numerical counts for high intensity pixels in the images. The research presents an automated approach for early detection of plant diseases using image processing techniques. By analyzing the color features of leaf areas, the k-means algorithm for color segmentation and the Gray-Level Co-Occurrence Matrix (GLCM) are used for disease classification. A novelty of this research is that it illustrates four categories of plants to analyze and compare: (1.) Grain, represented by Rice Plant Leaf Data; (2.) Fruit, represented by banana plant leaf data, (3.) Flower, represented by sunflower plant leaf data; and (4.) Vegetable, represented by potato plant leaf data. Six stages of image processing are applied to real data for diseases of leaf smut for rice, black sigatoka for banana, leaf scars for sunflower, and late blight for potato. Finally, a comparison of the image processing for each of the four plant types, conclusions, and future research directions are presented.
In the field of computer graphics, accurate representation of material properties is crucial for rendering realistic imagery. This paper focuses on the bidirectional reflectance distribution function (BRDF) and its role in determining how materials interact with light. The authors review the state of the art in reflectance measurement systems, with a focus on BRDF and bidirectional texture function (BTF) measurement. They discuss practical limitations in measuring multi-dimensional functions and provide examples of how researchers have addressed these challenges. Additionally, they analyse various approaches to converting measured data into practical analytical functions for use in commercial rendering software, including data-driven methods such as neural networks and hybridized approaches.
In this paper, Feature Selection Technique (FST) namely Particle Swarm Optimization (PSO) has been used. The filter based PSO is a search method with Correlation-based Feature Selection (CBFS) as a fitness function. The FST has two key goals of improving classification efficiency and reducing feature counts. Artificial Neural Network (ANN) Based Multilayer Perceptron Network (MLP) and Deep Learning (DL) have been considered the classification methods on 2 benchmark Autistic Spectrum Disorder (ASD) dataset. The experimental result was compared to the non-reduced features and reduced feature of ASD datasets. The reduced feature give up enhanced results in both classifiers MLP and DL. In addition, an experimental study on the exhibitions of these methodologies has been conducted. Finally, a new trend of PSO-MLP and PSO-DL based classification model is proposed.
In brain image classification, feature set reduction is essential to build an optimised feature subset that will lead to precise measurement. In this paper, an improved technique for feature selection by Moth Flame Optimization with Opposition Based Learning (OBL) and Simulated Annealing (OB-MFOSA) is proposed. The OBL strategy is used to create the optimum initial solution, while Simulated Annealing improves the search space. The proposed OB-MFOSA shows improved performance than other well-known existing algorithms by eliminating getting stuck in the local optima. By using this hybrid moth flame optimization, the feature set is reduced to 40%. Also, image denoising is performed by Dual Tree Complex Wavelet Transform (DTCWT) with an improved Log Gabor filtering technique. The filter bank of Log Gabor filter bank is tuned by Genetic Algorithm. The selected features from hybrid MFO algorithm are classified using SVM classifier. Experiments reveal that this hybrid algorithm shows accurate classification outputs than the previous methods.
Action Recognition plays a vital role in many secure applications. The objective of this paper is to identify actions more accurately. This paper focuses on the two stream network in which keyframe extraction method is utilized before extracting spatial features. The temporal features are extracted using Attentive Correlated Temporal Feature (ACTF) which uses Long Short Term Memory (LSTM) for deep features. The spatial and temporal features are fused and classified using multi Support Vector Machine (multiSVM) classifier. Experiments are done on HMDB51 and UCF101 datasets. The results of the proposed method are compared with recent methods in terms of accuracy. The proposed method is proved to work better than other methods by achieving an accuracy of 96% for HMDB51 dataset and 98% for UCF101 dataset.
In this paper, automatic and real-time systems were developed to detect and classify two different genera of scorpions using computer vision and deep learning techniques, with the purpose of providing a prevention tool. The images of scorpions were obtained from an arachnology laboratory in Argentina. YOLO (you only look once) and MobileNet models were implemented. The data augmentation technique was applied to significantly increase the amount of training data. High accuracy and recall values have been achieved for both models, which guarantees that they can early and successfully detect scorpions. In addition, the MobileNet model has shown to have excellent performance to detect scorpions within an uncontrolled environment, to carry out multiple detections, and to recognize their danger in case of accidents. Finally, a comparison has been made with other different machine learning-based models used to identify scorpions.
For accurately diagnosing the severity of brain tumors in MRI images, Glioma segmentation is a significant step. The Glioma segmentation is due to noise and weak edges of organs in medical images. The geodesic active contour model (GACM) is a standard method for the segmentation of complex organ structures based on edge maps. The GACM performs poorly due to this noise and weak edges. So, the authors propose a method that uses adaptive kernels instead of a constant kernel for creating strong edge maps for GACM. The kernels used in phase congruency are Log Gabor kernels, which resemble similar anisotropic properties like Gabor kernels. They have replaced these with adaptive kernels. This adaptive kernel-based phase congruency provides a robust edge map to be used in GACM. Experimentation shows that when compared with state-of-the-art edge detection techniques, adaptive kernels enhance the weak as well as strong edges and improve the overall performance. Moreover, the proposed methodology substantially requires fewer parameters compared to existing ACM methods.
Facial expression plays an important role in communicating emotions. In this paper, a robust method for recognizing facial expressions is proposed using the combination of appearance features. Traditionally, appearance features mainly divide any face image into regular matrices for the computation of facial expression recognition. However, in this paper, we have computed appearance features in specific regions by extracting facial components such as eyes, nose, mouth, and forehead, etc. The proposed approach mainly has five stages to detect facial expression viz. face detection and regions of interest extraction, feature extraction, pattern analysis using a local descriptor, the fusion of appearance features and finally classification using a Multiclass Support Vector Machine (MSVM). Results of the proposed method are compared with the earlier holistic representations for recognizing facial expressions, and it is found that the proposed method outperforms state-of-the-art methods.
Modern artificial intelligence systems have revolutionized approaches to scientific and technological challenges in a variety of fields, thus remarkable improvements in the quality of state-of-the-art computer vision and other techniques are observed; object tracking in video frames is a vital field of research that provides information about objects and their trajectories. This paper presents an object tracking method basing on optical flow generated between frames and a ConvNet method. Initially, optical center displacement is employed to detect possible the bounding box center of the tracked object. Then, CenterNet is used for object position correction. Given the initial set of points (i.e., bounding box) in first frame, the tracker tries to follow the motion of center of these points by looking at its direction of change in calculated optical flow with next frame, a correction mechanism takes place and waits for motions that surpass a correction threshold to launch position corrections.
Face Recognition is an efficient technique and one of the most liked biometric software application for the identification and verification of specific individual in a digital image by analysing and comparing patterns. This paper presents a survey on well-known techniques of face recognition. The primary goal of this review is to observe the performance of different face recognition algorithms such as SVM (Support Vector Machine), CNN (Convolutional Neural Network), Eigenface based algorithm, Gabor Wavelet, PCA (Principle Component Analysis) and HMM (Hidden Markov Model). It presents comparative analysis about the efficiency of each algorithm. This paper also figure out about various face recognition applications used in real world and face recognition challenges like Illumination Variation, Pose Variation, Occlusion, Expressions Variation, Low Resolution and Ageing in brief. Another interesting component covered in this paper is review of datasets available for face recognition. So, must needed survey of many recently introduced face recognition aspects and algorithms are presented.
Impulse and Gaussian are the two most common types of noise that affect digital images due to imperfections in the imaging process, compression, storage and communication. The conventional filtering approaches, however, reduce the image quality in terms of sharpness and resolution while suppressing the effects of noise. In this work, a machine learning-based filtering structure has been proposed preserves the image quality while effectively removing the noise. Specifically, a support vector machine classifier is employed to detect the type of noise affecting each pixel to select an appropriate filter. The choice of filters includes Median and Bilateral filters of different kernel sizes. The classifier is trained using example images with known noise parameters. The proposed filtering structure has been shown to perform better than the conventional approaches in terms of image quality metrics. Moreover, the design has been implemented as a hardware accelerator on an FPGA device using high-level synthesis tools.
Baby Sign Language is used by hearing parents to hearing infants as a preverbal communication which reduce frustration of parents and accelerated learning in babies, increases parent-child bonding, and lets babies communicate vital information, such as if they are hurt or hungry is known as a Baby Sign Language . In the current research work, a study of various existing sign language has been carried out as literature and then after realizing that there is no dataset available for Baby Sign Language, we have created a static dataset for 311 baby signs, which were classified using a MobileNet V1, pretrained Convolution Neural Network [CNN].The focus of the paper is to analyze the effect of Gradient Descent based optimizers, Adam and its variants, Rmsprop optimizers on fine-tuned pretrained CNN model MobileNet V1 that has been trained using customized dataset. The optimizers are used to train and test on MobileNet for 100 epochs on the dataset created for 311 baby Signs. These 10 optimizers Adadelta, Adam, Adamax, SGD, Adagrad, RMSProp were compared based on their processing time.
Detecting violence in real time videos is not an easy task even for the most advanced deep learning architectures, considering the subtle details of human behavior that differentiate an ordinary from a violent action. Even with the advances of deep learning, human activity recognition(HAR) in videos can only be achieved at a huge computational cost, most of the time also requiring special hardware for reaching an acceptable accuracy. We present in this paper a novice method for violence detection, a sub-area of HAR, which outperforms in speed and accuracy the state of the art methods. Our method is based on features extracted from the Pose estimator method OpenPose. These features are then transformed into more representative elements in the context of violence detection, which are then submitted to a LSTM neural network to learn how to identify violence. This work was inspired by the violencedetector.org, the first open source project for violence detection in real time videos.
Dynamic Adaptive Streaming over HTTP (DASH) is an emerging solution that aims to standardize existing proprietary streaming systems. DASH specification defines the media presentation description (MPD), which describes a list of available content, URL addresses, and the segment format. High bandwidth demands in interactive streaming applications pose challenges in efficiently utilizing the available bandwidth. In this paper, a novel Relative Strength Index (RSI) with Geometric mean (GM) namely RSI-GM is proposed for estimating available bandwidth for DASH. The proposed work starts by taking the video as an input at the transmitter side and then the video compression is performed using the TRLE. Then MD5 hashing-based AES encryption is applied to the compressed video data to provide data security. Then RSI-GM is proposed to estimate the available bandwidth for DASH. Finally, after estimation, the bitrate for estimated bandwidth is selected optimally using the Improved Shark Smell Optimization (ISSO) algorithm.
Emotion analysis is an area which is been widely used in the forensic crime detection domain, a mentoring device for depressed students, psychologically affected patient treatment. The current system helps only in identifying the emotions but not in identifying the level of emotions like whether the individual is truly happy/sad or pretending to be happy /sad. In this proposed work a novel methodology has been introduced. We have rebuilt the Traditional Local Binary Pattern (LBP) feature operator to image the expression and combine the abstract characteristics of facial expression learned from the neural network of deep convolution with the modified features of the texture of the LBP facial expression in the full connection layer. These extracted features have been subjected as input for CNN Alex Net to classify the level of emotions. The results obtained in this phase are used in the confusion matrix for analysis of grading of emotions like Grade-1, Grade-2, and Grade-3 obtained an accuracy of 87.58% in the comparative analysis.
This paper presents a study about screening large radiological image streams produced in hospitals for earlier detection of lung nodules. Being one of the most difficult classification tasks in the literature, our objective is to measure how well state-of-the-art classifiers can screen out the images stream to keep as many positive cases as possible in an output stream to be inspected by clinicians. We performed several experiments with different image resolutions and training datasets from different sources, always taking ResNet-152 as the base neural network. Results over existing datasets show that, contrary to other diseases like pneumonia, detecting nodules is a hard task when using only radiographies. Indeed, final diagnosis by clinicians is usually performed with much more precise images like computed tomographies.
One of the most serious global health threat is COVID-19 pandemic. The emphasis on improving diagnosis and increasing the diagnostic capability helps stopping its spread significantly. Therefore, to assist the radiologist or other medical professional to detect and identify the COVID-19 cases in the shortest possible time, we propose a computer-aided detection (CADe) system that uses the computed tomography (CT) scan images. This proposed boosted deep learning network (CLNet) is based on the implementation of Deep Learning (DL) networks as a complementary to the Compressive Learning (CL). We utilize our inception feature extraction technique in the measurement domain using CL to represent the data features into a new space with less dimensionality before accessing the Convolutional Neural Network. All original features have been contributed equally in the new space using a sensing matrix. Experiments performed on different compressed methods show promising results for COVID-19 detection. In addition, our novel weighted method based on different sensing matrices that used to capture boosted features demonstrates an improvement in the performance of the proposed method.
Sparse Representation (SR) and Dictionary Learning (DL) based Classifier have shown promising results in classification tasks, with impressive recognition rate on image data. In Video Semantic Analysis (VSA) however, the local structure of video data contains significant discriminative information required for classification. To the best of our knowledge, this has not been fully explored by recent DL-based approaches. Further, similar coding findings are not being realized from video features with the same video category. Based on the foregoing, a novel learning algorithm, Sparsity based Locality-Sensitive Discriminative Dictionary Learning (SLSDDL) for VSA is proposed in this paper. In the proposed algorithm, a discriminant loss function for the category based on sparse coding of the sparse coefficients is introduced into structure of Locality-Sensitive Dictionary Learning (LSDL) algorithm. Finally, the sparse coefficients for the testing video feature sample are solved by the optimized method of SLSDDL and the classification result for video semantic is obtained by minimizing the error between the original and reconstructed samples. The experimental results show that, the proposed SLSDDL significantly improves the performance of video semantic detection compared with state-of-the-art approaches. The proposed approach also shows robustness to diverse video environments, proving the universality of the novel approach.
In our everyday life, we come across various media that facilitate communication. Photography is one such medium used for visual communication. Although it is easy for human beings to look at a picture and describe it, it is often a hard task for a computer to generate a caption automatically if a photograph is fed to it. The recent development in deep learning and neural network has made this problem easier to work on especially if the relevant datasets are provided. This paper attempts to comprehensively summarize and present a unique perspective of the prevalent systems developed to address this problem of image captioning.
