Search

Walking has become a daily activity that we cannot miss, even walking can improve the health of our bodies. However, walking without proper procedures can result in repeated pressure on the plantar feet, leading to a high risk of foot ulcers. This study proposed a deep learning model to classify the walking duration and shoe pressure using plantar region pressure images. Convolutional Neural Network (CNN) is used in developing a shoe pressure classification model and walking duration with ResNet-50 architecture. This model uses plantar pressure images at the First toe (T1), First metatarsal head (M1), Second Metatarsal Head (M2), and Heel (HL) with three different shoe pressures (small at 80 mmHg, moderate at 160 mmHg, and high at 240 mmHg) and two walking duration (10 minutes and 20 minutes) from 16 patients (2304 images). From 16 patients, 13 were randomly selected to become training data, and the other three were chosen to become test data in evaluating the model's performance. The best accuracy was shown when classifying the walking procedure using 240mmHg shoe pressure with 10 minutes walking duration from the shoe pressure effect producing an F1-score of 0.89, and the walking duration effect shows an F1-score of 0.86 with T1 and M1 that tend to be easier to classify than M2 and HL. [ABSTRACT FROM AUTHOR]

World Health Organization (WHO) states that approximately 1.3 million of people die every year because of road traffic crashes. Microsleep is one of the factors of traffic crashes and the number of accidents caused by microsleep increase rapidly each day. This leads to a major accident resulting to higher number of deaths, injuries, demolition of properties and permanent disabilities. To overcome this situation, microsleep detection approach of smart vehicle needs to be studied in detail. Hence, in this project work, we use Convolution Neural Network (CNN) algorithm for this purpose. This proposed work implies face features as input data to evaluate the accuracy of microsleep traits while driving the smart vehicle. Face features data are capable to predict microsleep. CNN plays a significant role in this project due to its capability in performing both generative and descriptive tasks especially in image and video recognition. Thus, the model will analyze between open, closed, no yawn and yawn expression. This integration can be evaluated by detecting the change in face features from normal expression to microsleep symptoms such as opened eyes degree, closed eyes degree and mouth yawning. Results of the preliminary stages of this research are included and the input dataset is gained from Kaggle platform. The preliminary results show the microsleep model's prediction with 72% accuracy by using CNN algorithm. [ABSTRACT FROM AUTHOR]

We propose an unsupervised image segmentation approach, that combines a variational energy functional and deep con-volutional neural networks. The variational part is based on a recent multichannel multiphase Chan-Vese model, which is capable to extract useful information from multiple input images simultaneously. We implement a flexible multiclass segmentation method that divides a given image into K different regions. We use convolutional neural networks (CNNs) targeting a pre-decomposition of the image. By subsequently minimising the segmentation functional, the final segmentation is obtained in a fully unsupervised manner. Special emphasis is given to the extraction of informative feature maps serving as a starting point for the segmentation. The initial results indicate that the proposed method is able to decompose and segment the different regions of various types of images, such as texture and medical images and compare its performance with another multiphase segmentation method. [ABSTRACT FROM AUTHOR]

Malaria is an infectious disease caused by a parasite known as Plasmodium, which infects red blood cells. Malaria detection can be done by using a Blood Smear examination through a microscope. The patient's blood sample will be stained with fluorescence and then viewed using a microscope to see the presence of Plasmodium parasites in the blood. Computer vision and deep learning method can also be used to detect malaria from red blood cell images. Previous studies showed convolution neural network (CNN) model can accurately detect malaria from red blood cell images. However, the CNN architecture and parameters played an important role in increasing accuracy. This study investigated the effect of hyperparameter tuning toward accuracy, precision, recall, f1-score and Matthew Correlation Coefficient (MCC). Two architectures, Rajaraman and BaselineNet were used for the experiment and grid searching technique to find the best hyperparameter used in this research. The best hyperparameter for Rajaraman architecture were 30 epochs, ADAMAX optimizer, 32 batch-size, and 0.01 learning rate with 96.7% accuracy. The best hyperparameter for BaselineNet architecture were 20 epochs, SGD optimizer, 0.1 learning rate and 32 batch size. Overall, the hyperparameter tuning can increase 1.93% for Rajaraman architecture and 0.36% for BaselineNet architecture compared to non-tuning hyperparameter. [ABSTRACT FROM AUTHOR]

Birds have so many kinds of species, and a bird's attractiveness depends on its shape, colour, and whether it singing. Unfortunately, there are 177 bird species in Indonesia that will disappear because of poaching and illegal trade. With good information from the bird's picture, it can help people stop trading vanished birds. There are 15 different species birds, including javan sparrow, javanese eagle, parrot king ambon, golden julang, green peacock, rainbow lorikeet, katsuri ternate, black head katsuri, king cockatoo, yellow-crested cockatoo, tanau parrot, black-winged white starling, sepah raja honey, javanese glasses, and white-crested cockatoo. Each species of bird contains 200 pictures, which are split into 98.7% for data training and 1.3% for data testing. This study uses a convolution neural network with Adam Optimizer and Nadam Optimizer for bird classification and uses a confusion matrix for evaluation. The CNN with AlexNet Architecture has 97,14% in accuracy training, 91.62% in accuracy validation, and 0.3 loss validation for the training model with 100 epochs, a learning rate of 0.001, and Adam Optimizer. According to the confusion matrix results show that bird in fifth class can be recognized with true 31 pictures form 39 pictures. The CNN with AlexNet Architecture has 97.99% in accuracy training, 90.60% in accuracy validation, and 0.38 loss validation for the training model with 500 epochs, a learning rate of 0.0001, and Nadam Optimizer. According to the confusion matrix results show that bird in seventh class can be recognized with true 31 pictures form 39 pictures [ABSTRACT FROM AUTHOR]

In this study, we developed a new fuzzy logic-based convolution layer on a two-dimensional Convolutional Neural Network (2D-CNN). This innovation aims to enhance the ability of CNN in recognizing colors. We experimented on P3Net, which is a 2D-CNN model that is used to predict photosynthetic pigment content in plant leaves in real time and non-destructive manner. The P3Net is designed to be able to predict three main photosynthetic pigment content (chlorophyll, carotenoid, and anthocyanin) based on the leaves color. The leaf colors were captured in the form of an RGB image and the image was used as the CNN input. We compare the performance of P3Net with and without the fuzzy logic-based convolution layer. It was revealed that the new form of convolution layer could significantly improve the P3Net performance. [ABSTRACT FROM AUTHOR]

It takes time, money, and effort to keep a watch on your home. Numerous catastrophes, such as burglaries and vandalism, had a place in homes when the owners were negligent or absent. While employing staff is not thought to be a cost-effective alternative, some residential areas employ guards to watch over their homes. A mobile application-based Internet of Things (IoT) system called Home Intruder Detection System (HIDS) assists homeowners in keeping an eye on their houses by remotely alerting users to any potential dangers. The main goals of HIDS are to develop a trustworthy home security system using IoT, to apply the object detection algorithm to identify human presence, and to develop an intelligent mobile application that will allow users to monitor their homes from anywhere in the world and receive alerts if any threats are found. To identify intruders using a camera attached to the system, HIDS is used tested and compared the Region-based Convolution Neural Network (R-CNN), Single-Shot Multibox Detection (SSD) and You Only Look Once (YOLO) detection object detection method in the NVIDIA Jetson Nano. The system is capable of delivering detection video to the server and capturing video at frame rate (FPS) of 44.25 for YOLO, 37.81 for SSD and 27.21 for faster R-CNN respectively. Faster R-CNN, SSD, and YOLO algorithms' average precisions will be 94.08%, 89.75%, and 81.92% respectively. HIDS achieves its objectives by successfully recognizing people and remotely informing detection users via mobile applications. [ABSTRACT FROM AUTHOR]

Now a days we see more people are dying because of heart attacks and cardiac arrest. The Electrocardiogram is mostly used for identifying the heart diseases and also, we can find the cardiovascular state of the person. The deep learning is an approach to study about the abnormalities present in an ECG cycle. In this project we discuss about two methods reduction of noise and detection of abnormalities. In this approach we use Convolutional neural networks which gives the accuracy between 75 to 85% and also this method can identify the minute changes in the given signal. This process is done using MIT-BIH database. [ABSTRACT FROM AUTHOR]

Pose estimation is a fascinating aspect of pattern recognition that is used in a variety of industries, such as technology, healthcare, gaming, etc. The goal of human posture estimation is to foretell the positions of joints and body components in still photos and moving pictures. Given the wide range of uses for such a system, it constitutes one of the most intriguing domains of computer vision research and has had significant growth. The compact convolutional neural networks architecture for human posture estimation proposed in this study, called BlazePose, is intended towards mobile device legitimate prediction. During analysis, the network operates from over 30 fps on something like a Pixel 2 mobile, providing 33 body keypoints for such a specific individual. As a result, it is particularly well-suited for real-time applications like as fitness tracking and action recognition comprehension. This paper makes two major contributions: a novel body posture tracking approach and a lightweight body pose estimation computational model that evaluates parameters utilizing heatmaps and regression. The suggested method has improved estimate accuracy, which is useful when it is used for robotics, augmented reality, sports and fitness, action recognition, motion analysis, and human-computer interface. [ABSTRACT FROM AUTHOR]

An intelligent CSI feedback method is suggested for the fast time frequency conversion division duplex (FDD) large-scale multiple-input multiple-output (MIMO) system in wireless channel interference because the channel state information (CSI) matrix has noise and the Doppler frequency shift-induced time correlation prevents the system from providing high-reliability and low-latency communication. In order to extract noise from the CSI matrix and discover the spatial organisation of the channel, this technique combines convolutional neural networks (CNN) and batch normalisation (BN) networks. To increase the accuracy of the CSI reconstruction, the time correlation between the CSI matrices is reconstructed using the attention approach. The network is trained offline using the results of the standard fast time-varying channel model simulation. System modelling plus analysis show that the proposed method can successfully extract the Doppler-induced time correlation and limit the impact of noise. The proposed method performs better in terms of normalised mean square error (NMSE) and cosine similarity when compared to the representative CSI compressed feedback method and CsiNet method. [ABSTRACT FROM AUTHOR]

Skin disease is the most uncertain medical issue to detect. Mostly, it is extremely difficult to recognise skin disease with the naked eye. Before we go to a dermatologist, it is better to have self-care by knowing the kind. There are hundreds of categories of skin diseases. A few of them look very similar. We make wrong identification and implement incorrect measures that result in getting wrong results and also harming ourselves. So, to solve this problem, a new application was developed. It takes the image as input, and the name of the disease and control measures are displayed as output. The application takes the input image and processes it with the dataset already given and categorises the disease. The name of the disease and tips to reduce the spread of the disease are given to the user. Python code is used for the processing of images, and we employ a method called deep convolution neural networks (CNN). We are categorising the given image into four types: acne, benign, malignant, and skin allergy. The datasets are collected from different sources, such as Kaggle, GitHub, and Google. There are approximately 3003 training datasets and 600 test datasets in total. This network will give an accuracy of 98.7% and will produce results quicker than the formal technique, making this submission an effective and reliable system for dermatologic disease detection. Biomedical students in the dermatology branch can also use this as a consistent real-time training tool. [ABSTRACT FROM AUTHOR]

Motorcycle accidents increase due to high speed and traffic congestion. Riding motorcycles without wearing helmet is violating the rules and it may endanger the person's life. This paper aims to detect the motorcyclist failing to wear helmet and to recognize the vehicle license plate to display the number. The system is implemented using convention neural network (CNN) with you look only once (YOLO) architecture. [ABSTRACT FROM AUTHOR]

Wildlife photography is one of the best ways to connect with nature without disturbing its habitat and is a good way to convey the importance of nature. Wildlife photography is the best way to show how beautiful nature is and the creatures within nature. Wildlife photographers capture different species. They capture thousands and lacs of images of different species. They need help finding the photos of the specie they require. Here we are trying to classify the images into separate ones based on specie. They classify the images into four classes: animals, birds, snakes, and insects. Our world may be helpful for photographers to classify their images and have them into separate folders, so they can find them easily anytime they want. We are trying to develop an image classifier using artificial intelligence. As you know, artificial intelligence is one of the best ways to recognize things automatically. We implement it with the help of convolution neural networks along with the VGG 19 algorithm. [ABSTRACT FROM AUTHOR]

It is a known fact that any update in the history of educational sector has always been a positive impact in the livelihood of people towards technology. Our project is one such a kind where rating essays is the major criteria we want to work on. Essay evaluation is considered as a systematic way to give rating to the essays written. Automatic essay scoring is a process of grading essays without human intervention. The computer systems are trained using technical, artificial intelligence architectures where natural language processing comes into picture. The process of making machine resembles to the human intelligence and to work, as if as a human could is the main motive of natural language processing. Under this criterion, we have chosen a part of educational preview to build a system that is capable of rating written work, namely essays. Our project aims to provide a solution that evaluates essays as an automatic process. The basic idea here is to develop a software system that can be beneficial to educational institutions, business organizations, researchers, etc. Automatic essay scoring has a powerful gain over making it work, because it helps in reduction of manual work, gives a scope for every element without bias, also act as a key role in being time-efficient. There are past approaches in finding a way to develop an automated system to score essays using regression analysis, convolution neural networks, while we worked through transformer-based model, named BERT. [ABSTRACT FROM AUTHOR]

A human-assisted manual classification might result in erroneous prediction and diagnosis, brain tumor identification is one of the most important and difficult challenges in the realm of medical image processing. Moreover, when there is a big number of data to be helped, it is difficult to work. Because brain tumors have such a wide range of appearances and because tumor and normal tissues are so comparable, extracting tumor areas from pictures is difficult. We suggested a method for extracting brain tumors from "magnetic resonance brain images (MRI)" using a Convolution neural network algorithm, followed by standard "convolutional neural network" categorization. The experiment used a real-time dataset. "Neural Network (CNN)" is a type of neural network that is created using Kera's and Tensor Flow and outperforms regular neural networks. Information on abnormal tissue development in the brain can be gleaned from MRIscanning. [ABSTRACT FROM AUTHOR]

There are very few automobiles in developing countries because two wheelers have always been the uppermost mode of transport. Motorcycle crashes have been on the rise in the last few years. A number of people who are involved in traffic collisions include motorcyclists who do not wear reflective helmets, since they do not believe they provide sufficient protection. Once the traffic police spot those driving Motorcycles on a whole-or Motorcycles in junctions-without helmets, they also use video from CCTV to take control of the drivers of those vehicles and penal those who are riding without one. However, it can only be achieved through human action and commitment. Again they classifies moving vehicles as motorcycle or nonmotorcycle. for example, when referring to the head component, in the case of a motorcyclists, it is graded as either full face or non-full face. An excellent image analysis of the motorcycle number is then used to extract the characters that were missed by the identification software and/ Finally, the character count of the motorcycle is found, and from the motorcycle is examined using OCR software. The algorithm detects Objects to discover faces from still photos or videos. In their study "Rapid Object Detection using a Boosted Cascade of Simple Characteristics," Viola and Jones presented edge or line detection features, which are used by the technique. A deep learning algorithm called a convolutional neural network may analyse a picture as input and rank its significance. to separate elements/objects in the picture and be able to tell one from the other. Because of its great accuracy, CNNs are utilized for picture categorization and recognition. With the use of a hierarchical model, the CNN creates a network in the form of a funnel that eventually yields a fully connected layer in which all of the neurons are interconnected and the output is processed. [ABSTRACT FROM AUTHOR]

Mineralogy plays an important role in the early stages of earth resource exploration. In the past decades, geologists started to use hyperspectral in various scales of mineral identification, from hand specimen samples to airborne geological mapping surveys. Performing conventional mineral classification with hyperspectral methods is time-consuming and requires significant human intervention, from the endmember selection for each mineral to validating the classes manually in laboratory analysis. In this study, we introduced supervised machine learning algorithms to stimulate the mineral mapping process of a large dataset of core data. Three hyperspectral imageries of milled pebbled samples were used where one of the samples was pre-identified and used for training machine learning models to identify the mineralogy of the other two samples. The samples contain four minerals; namely Muscovite, Tourmaline, Illite and High-Crystallinity Illite, that will be auto-identified by the machine learning algorithms. In this study, Random Forest and Convolutional Neural Network were the selected algorithms to perform the mineral identification. Both algorithms produce high-accuracy mineral maps compared to the existing mineral maps from the previous study. The Convolutional Neural Network struggled to identify High-Crystallinity Illite, whilst Random Forest succeeded in separating High-Crystallinity Illite from other minerals. Thus, the Random Forest algorithm produces higher accuracy results. The proposed workflow provides a time-efficient alternative methodology for further mineral mapping process on a larger scale dataset. [ABSTRACT FROM AUTHOR]

Deep convolutional neural network (CNN) models were developed in this research using data from the Best Track of tropical storms, as well as ocean and atmospheric reanalysis, to forecast stimulated tropical cyclone intensity (TCI), minimum central pressure (MCP), and maximum two-min mean wind speed (MWS). Sensitivity tests were developed to assess the model structure's interpretability by testing various combinations of predictors. The results of the model test indicated that the simplified VGG-16 (VGG-16 s) model performed better than the other two general models (Le Net-5 and Alex Net). The sensitivity studies, which supported the premise and perceptions, validated the validity and reliability of the model. According to the research, the significance of predictors changed depending on the goal. The correlations between vertical wind shear speed (VWSS), temperature at 500 hPa (TEM_500), and sea surface temperature (SST) were quite high, while MCP was less significant for TCI. More significant than other variables for MWS and SST were the temperatures at 500 hPa (TEM_500) and 850 hPa (TEM_850). As per the study's conclusions, deep learning might be an alternative approach to conduct comprehensive, quantitative research on tropical cyclones. Deep learning and convolutional neural networks were used in the research to forecast tropical cyclones, and the interpretability of the results was examined. [ABSTRACT FROM AUTHOR]

Machine learning and deep learning are crucial to AI and the development of computers. Human effort may be decreased in understanding, learning, making predictions, and a lot more through the introduction of machine learning and deep learning. This article compares classifiers like convolution neural networks on the basis of performance, reliability, time, sensitiveness, positive productivity, and specificity with using different parameters with the classifiers. The handwritten elements (A to Z, a to z, and 0 to 9) from the famous EMNIST dataset are presented. From aspiring machine learning and deep learning beginners to seasoned practitioners, handwritten character recognition has become increasingly popular. A machine to comprehend and classify the photographs of the framework that is being developed contains handwritten character as alphabets (A-Z, a-z) and characters (0-9). The process through which a machine learns to recognize characters is known as character recognition system. Therefore, the main challenge would be whenever predicting each character according to similarities between individuals such as 1 and 7, 5 and 6, 3 and 8, N and Z, and a and o etc. When multiple people write the same character in a range of different handwritings, the problem is exacerbated. Lastly, the structure and appearance of the characters are affected as well by the individuality and variation of each pen's hand writing. [ABSTRACT FROM AUTHOR]

In the field of agriculture, there exists a conflict between animal and human, leading the human life in danger by losing a massive quantity of resources in the forest zone too. To address this issue, smart detection and repulsive system for animal intrusion is done using Deep Neural Network (DNN). Historically, this problem has been solved through both deadly (such as shooting or trapping) and non-lethal (such as educating the public) approaches (e.g., chemical repellents, electric fences). On the other hand, some of the more common approaches result in pollution that is harmful to humans and ungulates equally, while others are prohibitively expensive to maintain and provide only a minor degree of reliability and efficiency at best. The proposed research combines the use of AI Computer Vision with deep convolutional neural networks (DCNN) to detect and locate animals with the use of species-specific ultrasonic emission (i.e., each species of different types) to repel them. An ultrasonic signal tailored to the kind of animal detected is generated by the Animal Repelling Module and sent back to the camera via the edge-computing device. [ABSTRACT FROM AUTHOR]

Image classification plays a vital role in classifying different objects without the human supervision. In manual system of image classification is more time consuming process and it is not suitable for future automated systems. The lack of a completely automated, low-cost system for real-time picture categorization highlights the continued difficulty of the subject. In this work, we offer Multiple Fruit Maturity Stage Classification, a method for automatically categorizing the ripeness of various fruits using computer vision and deep learning techniques. Convolution neural network (CNN) is used to extract the appropriate features for accurate classification. When opposed to its forerunners, Convolution neural network's (CNN) key benefit is that it can discover crucial elements automatically, without human intervention. The system shows better accuracy for both test and validation. [ABSTRACT FROM AUTHOR]

Crop diseases present a substantial threat to worldwide food security, emphasizing the critical importance of timely and precise disease detection to mitigate yield losses and support sustainable agriculture. This study introduces an innovative approach that harnesses the power of Machine Learning (ML) and the Internet of Things (IoT) to revolutionize the detection of crop diseases. Within this framework, IoT sensors strategically deployed across agricultural fields gather real-time environmental data, while high-resolution crop images are captured using drones and on-site cameras. ML models, including Convolutional Neural Networks (CNNs), harness this data to enable early and accurate disease detection. By integrating IoT capabilities, we can correlate disease outbreaks with environmental factors, thereby enhancing the decision-making process. The results are promising, consistently demonstrating high accuracy in disease detection and the potential for reducing the need for chemical treatments. This research paper signifies a significant step towards promoting sustainable farming practices and enhancing global food security. [ABSTRACT FROM AUTHOR]

Automatic Speech Recognition (ASR) is an efficient method of converting speech signals into textual format and is often referred to as Speech-to-Text System (S2T). We present an End-to-End (E2E) Deep Learning based Continuous Speech Recognition system in Kannada using Convolutional Neural Network (CNN) and Bidirectional Gated Recurrent Units (Bi-GRU) as a variant of Recurrent Neural Network (RNN). The model was built using 120 hours of manually selected continuous Kannada speech data. Experimentation has been performed on two different deep learning-based models. The accuracy with lowest Character Error Rate (CER) of 18.62% was achieved. The outcome of this experiment has been quite encouraging for developing end-to-end automatic speech recognition systems. [ABSTRACT FROM AUTHOR]

Traffic forecasting and prediction are crucial in urban planning, transportation management, and decision-making. Traditional methods often struggle to capture traffic data's complex spatial dependencies and temporal dynamics. In recent years, Graph Convolutional Neural Networks (GCNs) have emerged as a powerful tool for addressing these challenges. GCNs leverage graph structures to model relationships between traffic nodes, such as road intersections or segments, and have shown promising results in various traffic forecasting and prediction tasks. Within this paper, an extensive exploration unfolds, delving into the utilization of GCNs in the realm of traffic forecasting and prediction. It meticulously illuminates their inherent strengths, navigates through limitations, and envisions potential future trajectories. Additionally, a pioneering framework for traffic forecasting takes centre stage within these pages. [ABSTRACT FROM AUTHOR]

Particle identification (PID) is one of the most fundamental tools in various physics research conducted in collider experiments. In recent years, machine learning methods have gradually become one of the mainstream methods in the PID field of high-energy physics experiments, often providing superior performance. The emergence of quantum machine learning may potential arm a powerful new toolbox for machine learning. In this work, targeting at the π±/K± discrimination problem at the STCF experiment, a convolutional neural network (CNN) in the endcap PID system is developed. By combining the hit position and arrival time of each Cherenkov photon at the sensors, a two-dimensional pixel map is constructed as the CNN input. The preliminary results show that the CNN model has a promising performance. In addition, based on the classical CNN, a quantum convolution neural network (QCNN) is developed as well, exploring possible quantum advantages provided by quantum machine learning methods. [ABSTRACT FROM AUTHOR]

We are studying the use of deep neural networks (DNNs) to identify and locate primary vertices (PVs) in proton-proton collisions at the LHC. Earlier work focused on finding primary vertices in simulated LHCb data using a hybrid approach that started with kernel density estimators (KDEs) derived heuristically from the ensemble of charged track parameters and predicted "target histogram" proxies, from which the actual PV positions are extracted. We have recently demonstrated that using a UNet architecture performs indistinguishably from a "flat" convolutional neural network model. We have developed an "end-to-end" tracks-to-hist DNN that predicts target histograms directly from track parameters using simulated LHCb data that provides better performance (a lower false positive rate for the same high efficiency) than the best KDE-tohists model studied. This DNN also provides better efficiency than the default heuristic algorithm for the same low false positive rate. "Quantization" of this model, using FP16 rather than FP32 arithmetic, degrades its performance minimally. Reducing the number of UNet channels degrades performance more substantially. We have demonstrated that the KDE-to-hists algorithm developed for LHCb data can be adapted to ATLAS and ACTS data using two variations of the UNet architecture. Within ATLAS/ACTS, these algorithms have been validated against the standard vertex finder algorithm. Both variations produce PVfinding efficiencies similar to that of the standard algorithm and vertex-vertex separation resolutions that are significantly better. [ABSTRACT FROM AUTHOR]

Fast, efficient and accurate triggers are a critical requirement for modern high energy physics experiments given the increasingly large quantities of data that they produce. The CEBAF Large Acceptance Spectrometer (CLAS12) employs a highly efficient electron trigger to filter the amount of data recorded by requiring at least one electron candidate in each event, at the cost of a low purity in electron identification. However, machine learning algorithms are increasingly employed for classification tasks such as particle identification due to their high accuracy and fast processing times. In this proceeding we present recently published work that showed how a convolutional neural network could be deployed as a Level 3 electron trigger at CLAS12. We demonstrate that this AI trigger would achieve a significant data reduction compared to the conventional CLAS12 electron trigger, whilst preserving a 99.5% electron identification efficiency, at nominal CLAS12 beam currents. [ABSTRACT FROM AUTHOR]

Rainbow antimagic coloring is a research topic on graphs that can be applied in everyday life. The graph neural network itself is a deep learning architecture that is used to solve machine learning problems on data in the form of graphs. Graph Neural Networks (GNNs) are a topic that researchers are currently discussing. GNNs are capable of performing tasks that Convolutional Neural Networks cannot do. By looking at existing literature, this research will examine the use of Rainbow Antimagic Coloring of Graph and Graph Neural Networks. The integration of Rainbow Antimagic Coloring of Graph and Graph Neural Networks is expected to be able to solve problems in life. Based on the results obtained from existing literature, the rainbow antimagic coloring concept can be combined with the neural network concept to solve real-life problems, including irrigation systems and supply chain management. [ABSTRACT FROM AUTHOR]

In this article, a method has been proposed to predict stroke on earlier basis to prevent any critical damage or death caused by it. The aim is to predict stroke in earlier basis so the necessary steps can be taken to stave off any critical damage. For this purpose, the parameters like heart disease, smoking habit, hypertension, glucose level, body mass index level, age, and work pressure have been considered. Moreover, previously happened stroke is also being considered to observe and empathize the situations and based on these facts real time stroke prediction analysis using deep learning and machine learning approaches has been suggested. At the present time "machine learning" and "deep learning" tactics are put on for finding the problems' solutions in "computer vision", forecasting and "convolutional neural network". This work proposes a before time prediction of stroke using various machine learning methodologies with occurrences of parameters based on which different classifiers are trained and data processing is done which attains a better outcome than other methodologies. [ABSTRACT FROM AUTHOR]

Driver rest is one of the main sources of mishaps and passing for some individuals and the economy. As of late, specialists have started to perceive that lack of sleep is a positive, industry-driving reaction. In this article, we suggest including CNN-based savvy glasses for eye acknowledgment and rest estimation. This new response is contrasted with the customary technique in view of the strategy for comprehension. Both execution, battery utilization, and impression are assessed for execution in incorporated glass. The outcomes show that CNN has over 7% of the precision found at the algorithmic level. Moreover, migraines in memory and battery duration are genuine and make CNN a potential answer for sleep deprivation. [ABSTRACT FROM AUTHOR]

Pineapple diseases are linked to worms, viruses, bacteria, and fungi. Insects including ants, scales, mealybugs, and souring beetles are some of the most prevalent pests that can harm the pineapple industry over time if early pineapple leaf disease detection is not made There are many smallholders especially in the rural area who lack of latest technology, knowledge, and experience about pineapple disease treatment and management. Thus, lack of a precise diagnosis method prevents them from understanding the disease and creates an effective control for its spread. This issue further impacts the effectiveness of pineapple production, subsequently affecting their income and the country's, as well as jeopardizing national food security. This study presents an innovative approach aimed at enhancing early detection and management of pineapple leaf diseases by integrating a Convolutional Neural Network (CNN) algorithm and import it using TensorFlow Lite into an android mobile. The CNN model achieved an impressive total accuracy of 98% in precisely classifying three types of pineapple leaf diseases: Leaf Spot, Mealybug Wilt, and Pink Disease. When implemented in the mobile application, the system attained an overall confidence of 83.33% by leveraging both camera-captured and gallery images. Although the accuracy gained is impressive, additional research and adjustments can be conducted to improve the system's performance and achieve even higher levels of accuracy. [ABSTRACT FROM AUTHOR]

At present, detection and recognition of animals remain a complicated task and no distinctive technique can solve all situations strongly and effectively. Animal detection in images is challenging because of the complexity of wild environments. Including several different things: The large differences in shapes and color appearances of variations of objects which belong to the same class, and Lighting/Illumination Conditions. This paper presents an effective and efficient animal detection system for Birds detection in images. MobileNet SSD model is a Single Shot Multibox Detection (SSD) suggested in this work, which is capable of handling different shapes and view angles of the objects. This approach allows extracting features of image regions and using deep learning techniques for classifying regions into two categories: (bird and non-animal). The experimental results demonstrated the efficiency of classification and detection of single and multi-animals even in difficult environmental conditions. The accuracy of detection and classification is up to 98.8%. This methodology is more accurate than other similar approaches. This paper exhibits a method for training a convolutional neural network (CNN) based on object detection classifiers. [ABSTRACT FROM AUTHOR]

In many applications like human-computer interaction and video analytics, facial expression analysis has become one of the prime objectives. A simple approach for finding the comfortable level/happy mindset of a human is to recognize their smile. Hence, in this paper, we proposed an automatic smile recognition technique with high-level features extracted using convolutional neural networks. Here cascaded convolutional network model with YOLO V4 and LeNet-based architecture is used to detect the person, and, smile recognition. The experimental results of our proposed system provide higher accuracy of 96.82% than the existing techniques. [ABSTRACT FROM AUTHOR]

Brain Cancer is the most dangerous type of cancer, which affects hundreds of thousands of people worldwide every year. Early identification of brain tumor is essential for timely treatment, which is crucial for the patient's survival. Researchers have found it challenging to develop Computer-Aided Diagnosis (CAD) for Brain Tumor Identification. Many studies have discussed the feasibility of Deep Convolutional Neural Network algorithms like ResNet, Inception, VGG in Brain Tumor Classification and Detection. These Deep Learning algorithms seem to have struggled from vanishing gradient, model bias with overfitting due to smaller medical datasets available. The proposed model TumorNet adopts a custom activated Deep CNN Architecture for classifying the more dangerous malignant tumors from benevolent tumors. The proposed model is trained from scratch and able to train with a smaller dataset consisting of MRI scans. It is evaluated, and it gives an impressive accuracy of 98.24%, outperforming other state-of-the-art CNN models, ResNet-101 and EfficientNet-B0. In comparison, the F1 scores of the proposed model are estimated at 98.05%, significantly higher than the other two models, which are 82.46 and 88.33%, respectively. [ABSTRACT FROM AUTHOR]

The goal of the research is detection of disease in fresh fruits. The sensitivity and accuracy was calculated and compared with the k-nearest neighbor (KNN) classifiers. Materials and Methods: The research's data set was obtained from the Kaggle open access dataset. The research includes two groups, in which CNN is group 1 and KNN is group 2. Kaggle database is used for collecting different fruit images. The sample size was calculated using G power (80%) and alpha (0.05) and it is found to be 20 samples for each group. Result : CNN has achieved 98 % of accuracy and 1.0477 standard deviation while KNN method has achieved 90 % of accuracy and 0.65071 standard deviation. The Novel CNN has significant accuracy of p<0.05 (2 tailed). Conclusion: It is concluded that the detection of fruit disease using Convolution Neural Network appears to be significantly better compared to k-Nearest Neighbour Classifiers. [ABSTRACT FROM AUTHOR]

This paper's objective is to improve accuracy in plant leaf disease detection using Machine learning techniques such as K-Nearest Neighbors classifier (KNN) with a sample size of 10 and Novel Convolution Neural Network (CNN). This article consists of 2 groupsi.e Convolutional Neural Network (CNN) and K-Nearest Neighbors Classifier with a sample size of 10 for each group. G Power software is used to determine sample size with a pretest power value is 0.8 and alpha 0.05. The accuracy evaluated using both algorithms is compared and CNN appears to be the most significant with an accuracy of 92.48% and KNN with an accuracy of 74.14% with p=0.041. An Independent sample t-test has been used and analyzed with a confidence level of 95 percent. Results show that the CNN algorithm is significantly better than KNN in detecting plant leaf disease. [ABSTRACT FROM AUTHOR]

The primary objective of this study is to evaluate the efficacy of the Blaze Pose Algorithm, a new method for human pose detection, in comparison to that of Convolutional Neural Networks. Unlike the current method, which relies on static neural networks, Blazepose is a novel deep learning model for estimating human poses. Pretest power is 0.8, and there are 163 participants in each group. Data was gathered from the MPII Human Pose Dataset, which contains over 25,000 photos taken from various web videos. Researchers found that Blazepose's accuracy was slightly higher than that of a traditional neural network (90.3% vs. 40%). Independent sample T-test results show a 95% level of confidence that there is a statistically significant difference between the proposed and existing groups, with a significance value of P = 0.001 (P 0.05) 2-tailed. The Blazepose outperforms the traditional neural network in terms of improved accuracy, as seen by the comparison findings. [ABSTRACT FROM AUTHOR]

The purpose of this research is to use Machine Learning in Google COLAB to categorise human facial expressions. Images labelled as "happy," "sad," or "neutral" using a novel Support Vector Machine dataset. About 100 were selected for use in preparing the samples. As a training dataset, we used 25 pictures of people smiling broadly. The training dataset consists of 25 pictures of the expression "sad." The dataset used for testing included 25 images labelled with the "sad" emotion, and a similar number labelled with the "happy" emotion. The values of Gpower (80%) and alpha (0.5) were derived from SPSS. The mean difference for SVM in SPSS is 91.6667, while the mean difference for CNN is 87.3333. SVM has a variance of 2.16025 and CNN has a variance of 2.87228. When it comes to facial expression detection and classification, the proposed novel Support Vector Machine classifier has a 98 percent accuracy, while CNN only manages an 88 percent success rate. Regardless of the assumption of equal variances, a 2-tailed significance level of 0.002 is considered accurate. Mask-R Convolutional Neural Network has a 2-tailed significance of 0.840, which holds true even when assuming unequal variances. SPSS simulation yields the equal variances assumptions. The proposed novel Support Vector Machine outperforms CNN in Facial Expression Recognition, but only by a small margin. This demonstrates SVM's innovative potential in practical applications. [ABSTRACT FROM AUTHOR]

A comparative analysis of a novel convolutional neural network and k-nearest neighbors is performed to identify fingerprints in low-resolution photographs with high accuracy and sensitivity. Materials and methods: In this study, two groups were compared: a novel convolutional neural network (N=10) and a k-nearest neighbor algorithm (N = 10). Total sample size was estimated with an alpha of 0.05, an enrollment rate of 0.1, a confidence interval of 95%, and a power of 80% using G Power software. Results: The proposed method shows that the novel convolutional neural network achieves 93% accuracy and 90% sensitivity, while the K-Nearest Neighbor classifier achieves 89% accuracy and 87% sensitivity. The obtained accuracy rate was significant (p=0.005), and in the SPSS statistical analysis, the specificity value was (p=0.006). Conclusion: The new CNN classifier provides significantly better fingerprinting results than the K-Nearest Neighbors classifier. [ABSTRACT FROM AUTHOR]

The objective of this work is to increase the accuracy of credit card fraud prediction using the Support vector machine approach. Materials and Procedures: A G-power value of 89 percent was applied to iterations of a Support Vector Machine and a Convolutional Neural Network (CNN) with 100 samples each, in an innovative method for forecasting the total amount of credit card fraud that will be committed. Support Vector Machines are far more accurate than Convolutional Neural Networks, with a 98.2% accuracy rate (92.4). K-means clustering outperformed Convolutional neural network at the p=0.004 and p=0.005 levels (CNN). In conclusion, the support vector machine approach enables a state-of-the-art decision support system to reliably predict a broader spectrum of financial transactions. [ABSTRACT FROM AUTHOR]

Comparing the accuracy of novel VGGNet and extreme learning machine convolutional neural networks in estimating age and gender to optimize advertising for all audiences. The goal of this work is to improve the accuracy of age and gender recognition. In this work, an extreme learning machine convolutional neural network with a sample size of 10 and a VGGNet with a sample size of 10 are used to identify age and gender with a 95% confidence interval and 80% pretest performance. Targeted advertisements will be shown to you based on your recorded age and gender. The prediction accuracy rate of the VGGNets method reaches 93.90%, and the prediction accuracy rate of the extreme learning machine convolutional neural network method reaches 83.20%. Statistical analysis yielded a significant value of p=0.0298 (p0.5). The results show that VGGNets perform much better than extreme learning machine convolutional neural networks in recognizing age and gender. [ABSTRACT FROM AUTHOR]

The main objective of this research is to enhance the accuracy of plant disease classification by processing leaf images. This article consists of 2 groupsi.e Convolutional Neural Network (CNN) and Fuzzy Classifier with a sample size of 10 for each group. G Power software is used to determine sample size with a pretest power value is 0.8 and alpha 0.05. The Novel Convolution Neural Network and the Fuzzy algorithms were implemented and compared accuracy results. Convolution Neural Network appears to be more significant with 92.48% accuracy than Fuzzy Classifier with 82%. The CNN map appears to perform significantly better than the Fuzzy with the value of p=0.19. The result shows that the Convolution Neural Network algorithm's accuracy was better than other machine learning algorithms in Plant disease classification. [ABSTRACT FROM AUTHOR]

The study's goal is to create an efficient feedback system for automated and unmanned restaurants that makes use of customers' facial expressions by training the model with six distinct emotions. Substances and Techniques: To identify human emotions in cashierless eateries, the proposed system employs a unique gradient descent technique to the Improved support vector machine. The FER2013 dataset was utilised for this implementation. In this research, 392 people make up each of two groups. The outcomes showed that the proposed method was 82.14 percent accurate, which is a significant improvement over the current system's accuracy of 60.84 percent. Using a novel gradient descent technique, the results reveal that the upgraded support vector machine surpasses the current system in terms of accuracy when applied to the feedback system of unmanned and automated restaurants. This system can also be utilised as a resort feedback system to enhance the vacation experience for guests. [ABSTRACT FROM AUTHOR]

Convolution neural networks are being used in the study to accurately detect brain tumours, and the expectation accuracy and fuzzy C-Mean method are being compared. In the current framework Fuzzy C-Mean calculation is utilized and in proposed framework Convolution neural organizations is used. Fuzzy C-Mean algorithm (N=10) and CNN (N=10) iterated multiple times to foresee Brain Tumor disease.CNN has essentially improved precision (98.7%) contrasted with FCM exactness (90.5%). The significance value is 0.001 for CNN (p<0.05 free example test) is high. Within the restrictions of this examination, CNN is more precise in location of Brain Tumor sickness. [ABSTRACT FROM AUTHOR]

There are a number of technologies that might help farmers increase yields and decrease losses, but two of the most important are precision farming and accurate weather forecasts. Despite this, these systems need precise, up-to-date information on a wide range of factors, including soil quality, weather forecasts, and crop health. When it comes to potential destruction, cyclones are among the most concerning, and they may also have a major effect on farming methods. We can get a deeper understanding of the correlation between cyclones and crop yields by building and testing a custom-made convolutional neural network (CNN) using historical cyclone data. Eventually, this data might be utilised to create more accurate weather forecasts and precision farming algorithms to benefit farmers. Using this data, we can develop these cutting-edge algorithms. To train and verify the proposed framework, we used an existing dataset that includes cyclone data from the past. Compared to state-of-the-art approaches, the suggested architecture outperformed them significantly in terms of accuracy and precision. [ABSTRACT FROM AUTHOR]

The motive of the study is to reduce electricity loss of individual customers which is happening through power thefting by another customer and improve the accuracy of power theft detection using CNN-based LSTM algorithm. This approach was intended to find out the customers of power thefting. The data was collected by preprocessing data to train the model. This preprocessing step has been binded with synthetic data generation for better performance of the model. Later this trained model was hypertuned and implemented via the test data. There are two groups in this study each with a sample size of 21400 per group. This analysis has been done with the pretest power 0.8. The mean value for accuracy in CNN-LSTM method is 89.7735 which is high when compared to Zigbee whose accuracy is 81.8420. The statistical t test-difference between two independent means is 0.05(p<0.05). The study shows that the novel power theft detection technique in CNN-LSTM algorithm has more significant and reliable accuracy of power theft detection compared to the ZIGBEE model. [ABSTRACT FROM AUTHOR]