Journal of Intelligent Systems and Internet of Things

Script identification is crucial for document analysis and optical character recognition (OCR). This study proposes YafNet, a novel convolutional neural network (CNN) architecture, developed from scratch, to tackle the challenges of script identification in both handwritten and printed word images. YafNet dynamically weights features, enabling it to learn and combine multimodal features for accurate script identification. To evaluate its efficacy, we use the imbalanced ICDAR 2021 Script Identification in the Wild (SIW 2021) competition dataset. Experimental results demonstrate that YafNet outperforms conventional approaches, particularly when trained on mixed handwritten and printed data. It achieves high classification accuracy, balanced accuracy, and ROC AUC scores, indicating its robustness and generalizability. The incorporation of data augmentation and external data further enhances performance, underscoring the model's potential for real-world applications.

In current scenario, plant health monitoring plays a crucial role in effective health maintenance of plants in climate changes. Internet of Things (IoT) played an efficient role in realizing the remote and real-time monitoring of any physical things and activities through internet connectivity. In this study we have proposed a system that is able to monitor the plant health with the assimilation of wireless sensors and wireless network. The proposed system is able to log the sensor values on the plants on the cloud server through internet connectivity.

Predicting load demand is relevant when used in microgrid energy management systems to address issues such as nonlinear and dynamic consumption data. In this research, the author presents a fusion of Adaptive Sunflower Optimization (ASFO) and Enhanced Support Vector Machine (ESVM) methods to predict the load demand in micro grid environment. The ASFO algorithm enhances the efficiency of the ESVM through a fine-tuning meta-heuristic algorithm based on the sunflower natural organisms. This integration of ASFO and ESVM eliminates many of the drawbacks associated with the basic performance of the task, namely low speed of convergence, overtraining, and the presence of local minima in choosing the parameters. Some of the general parameters used in training and validating the model include load and meteorological data features involving, weather, temporal, load histories are the main contributors in the analysis. Comparisons with other ML algorithm ‘shave been made in respect of relative performance against established methods, such as Random Forest (RF) and Particle Swarm Optimization based with ESVM (PSO-ESVM). The findings infer that lower values of Mean Absolute Percentage Error (MAPE) and Symmetric Mean Absolute Percentage Error (SMAPE) and higher consistency index (d) are yielded by the proposed hybrid ASFO-ESVM model. For instance, even on working days of the week, the precision of the load forecasts was higher with the hybrid model than with the other options. The outcomes do prove that the proposed ASFO-ESVM model is very reliable and precise in its concerning aspect of load demand forecasting as it can be seen in the results obtained for different situations. Relatively, this work estimates a cost effective and feasible method for micro grid energy predictions which can enhance decisions in matters concerning power production, distribution, and control of energy. The study shows how these techniques are relevant towards the complexity and dynamism of the contemporary energy systems.

With an average annual intake of almost 20 kilograms per person, tomatoes are the most consumed vegetable worldwide. Diseases brought on by dangerous organisms are among the most important factors adversely affecting tomato production's output and quality. Depending on the climate and environmental conditions, tomatoes can be afflicted by a variety of illnesses throughout the planting and growing phases. It is essential for tomato growers to identify possible infections and take the appropriate preventative measures. Applications of artificial intelligence have grown in popularity recently. AI is being used in agriculture to identify plant illnesses. This research uses deep learning, a branch of artificial intelligence, to categories common tomato diseases. In the beginning, samples of frequently seen tomato illnesses were gathered from tomato growers in Kirkuk. Once there were enough data, the system developed with image processing algorithms produced meaningful images. Using a CNN-based GoogleNet deep learning system, the resulting dataset was trained and diseases were classified. The results show that the deep learning system that was constructed has a high degree of success and dependability when it comes to tomato disease classification.

Cotton is the most significant cash crop in India. Each year cotton production is decreasing because of the attack of the disease. Plant diseases are usually produced by pathogens and pest insects and reduce the yield to a large scale if not controlled in time. The hour requires an effective plant disease diagnosis system that can assist the farmers in their farming and cultivation. Nevertheless, cotton production is harmfully affected by the presence of viruses, pests, bacterial pathogens, and so on. For the past decade or so, numerous image processing or deep learning (DL)--based automated plant leaf disease recognition techniques have been established but, unluckily, they infrequently focus on the cotton leaf diseases. Therefore, this article develops an Intelligent Detection and Classification of Cotton Leaf Diseases Using Transfer Learning and the Honey Badger Algorithm (IDCCLD-TLHBA) model with Satellite Images. The proposed IDCCLD-TLHBA technique intends to determine and classify various kinds of cotton leaf diseases using satellite imagery. In the IDCCLD-TLHBA technique, the wiener filtering (WF) model is used to reduce noise and enhance image quality for subsequent analysis. For feature extraction, the IDCCLD-TLHBA technique applies the MobileNetV2 model to capture relevant features from the satellite images while maintaining computational efficiency. In addition, the stacked long short-term memory (SLSTM) method is employed for the classification and recognition of cotton leaf diseases. Eventually, the honey badger algorithm (HBA) is used to optimally select the parameters involved in the SLSTM model to ensure a better configuration of the network to enhance results. The performance validation of the IDCCLD-TLHBA method is carried out against the benchmark dataset and the stimulated results highlight the better results of the IDCCLD-TLHBA model across the existing techniques.

The demand for high-quality Direct-to-Home (D2H) television broadcasting services delivered via Low Earth Orbit (LEO) satellite constellations has surged in recent years. To address the growing needs of viewers, satellite communication must optimize the scheduling and routing of signals while balancing conflicting objectives. This research presents a novel approach named as Multi-Objective Deep Routing and Scheduling (MODRS) algorithm that is designed to tackle the challenges of signal latency minimization, bandwidth utilization maximization, and viewer demand satisfaction. The Multi-Objective Deep Neural Network (MODNN) is implemented in this paper to make intelligent routing and scheduling decisions for balancing multiple objectives. To enhance the learning process and provide training stability, the experience replay is used and the epsilon-greedy strategy is included to balance exploitation and exploration strategies. The Pareto-front concept is used for efficient D2H television broadcasting in the LEO satellite constellation. The experimental validation is conducted based on low-latency broadcasting, high-bandwidth utilization, viewer demand flexibility, adaptive signal strength and resource allocation efficiency. Using a series of simulated scenarios, this paper explores the versatility and robustness of MODRS, showcasing its exceptional performance in real-time, resource-efficient, disaster recovery, and rural broadcasting contexts. The findings indicate that MODRS is well-suited for a wide range of real-world applications, from low-latency broadcasting and disaster recovery to cost-effective rural expansion, enhancing the quality and accessibility of D2H television services. The MODRS algorithm emerges as a transformative solution for satellite communication optimization, ensuring viewer satisfaction and operational efficiency.

Machine learning (ML)-based forecasting techniques have demonstrated significant value in predicting postoperative outcomes, aiding in improved decision-making for future tasks. ML algorithms have already been applied in various fields where identifying and ranking risk variables are essential. To address forecasting challenges, a wide range of predictive techniques is commonly employed. Research indicates that ML-based models can accurately predict the impact of COVID-19 on Jordan's healthcare system, a concern now recognized as a potential global health threat. Specifically, to determine COVID-19 risk classifications, this study utilized three widely adopted forecasting models: support vector machine (SVM), least absolute shrinkage and selection operator (LASSO), and linear regression (LR). The findings reveal that applying these techniques in the current COVID-19 outbreak scenario is a viable approach. Results indicate that LR outperforms all other models tested in accurately forecasting death rates, recovery rates, and newly reported cases, with LASSO following closely. However, based on the available data, SVM exhibits lower performance across all predictive scenarios.

The Internet of Things (IoT) has significantly transformed the tourism industry, reshaping travel design, supply, and experiences. This paper reviews the key developments in tourism IoT from the mid-2010s, highlighting technological, economic, and socio-cultural impacts. It explores the adoption of IoT technologies –such as smart wearables, intelligent transportation systems, and augmented reality –across tourism sectors, emphasizing their effects on tourist behaviour and sustainable tourism development. A mixed-method approach, including literature reviews and expert interviews, is used to analyse these trends. Findings reveal that IoT enhances personalization, immersion, and sustainability in travel experiences, though privacy, security, and ethical issues pose challenges. Strategic planning and collaboration are necessary to leverage IoT innovations for sustainable tourism growth.

With the continuous progress of image retrieval technology, the speed of searching for the required image from a large amount of image data has become an important issue. Convolutional neural networks (CNNs) have been used in image retrieval. However, many image retrieval systems based on CNNs have poor ability to express image features. Content-based Image Retrieval (CBIR) is a method of finding desired images from image databases. However, CBIR suffers from lower accuracy in retrieving images from large-scale image databases. In this paper, the proposed system is an improvement of the convolutional neural network for greater accuracy and a machine learning tool that can be used for automatic image retrieval. It includes two phases; the first phase (offline processing) consist of two stages; stage1 for CNN model classification while stage 2 for extracts high-level features directly from CNN by a flattening layer, which will be stored into a vector. In the second phase (online processing), the retrieval depends on query by image (QBI) from the system, which relies on the online CNN model stage to extract the features of the transmitted image. Afterward, an evaluation is conducted between the extracted features and the features that were previously stored by employing the Hamming distance to return all similar images. Last, it retrieves all the images and sends them to the system. Classification for images was achieved with 97.94% deep learning results, while for retrieved images, the deep learning was 98.94%. For this paper, work done on COREL image dataset. The images in the dataset used for training are more difficult than image classification due to the need for more computational resources. In the experimental part, training images using CNN achieved high accuracy, proving that the model has high accuracy in image retrieval.

Gesture recognition is employed in human-machine communications, enhancing human life with impairments or who depend on non-verbal instructions. Hand gestures role an important role in the field of assistive technology for persons with visual impairments, whereas an optimum user communication design is of major importance. Many authors with substantial development for gesture recognition modeled several methods by using deep learning (DL) methods. This article introduces a Robust Gesture Sign Language Recognition Using Chicken Earthworm Optimization with Deep Learning (RSLR-CEWODL) approach. The projected RSLR-CEWODL algorithm majorly focuses on the recognition and classification of sign language. To accomplish this, the presented RSLR-CEWODL technique utilizes a residual network (ResNet-101) model for feature extraction. For optimal hyper parameter tuning process, the presented RSLR-CEWODL algorithm exploits the CEWO algorithm. Besides, the RSLR-CEWODL technique uses a whale optimization algorithm (WOA) with deep belief network (DBN) method for the sign language recognition method. The simulation result of the RSLR-CEWODL algorithm is tested using sign language datasets and the outcome was measured under various measures. The simulation values demonstrated the enhancements of the RSLR-CEWODL technique over other methodologies.

As online education has become increasingly prominent, the primary objective of this study is to evaluate students' opinions of online classes taught by teachers with no prior experience in online teaching, focusing on their teaching style, teaching efficiency, and pedagogy in the online classroom. Online teaching is a kind of teaching system that depends on network management technology. It concludes the teaching method by the process of live courses or recorded courses employing software containing special online teaching environments and any APP software employed for teaching. Social media, with its massive pool of user-generated content and instant feedback, offers a great opportunity to calculate teaching styles in online class management. Therefore, this study offers a Social Media Based Evaluation of Teaching Style in Online Education Systems using Heuristic Search (SMBETS-OESHS) Algorithm. The main objective of the SMBETS-OESHS technique for evaluate teaching styles in online education systems using insights derived from social media platforms. At primary stage, the SMBETS-OESHS model takes place linear scaling normalization (LSN) is implemented for scaling the input data. Next, the bayesian optimization algorithm (BOA) based feature selection process can be employed to allow for the detection of the most relevant features from the data. In addition, the SMBETS-OESHS model exploits stacked sparse autoencoder (SSAE) technique for classification process. In order to achieve optimal performance, the SSAE model parameters are fine-tuned using the improved beetle optimization algorithm (IBOA), ensuring robust evaluation accuracy. The experimental validation outcome of the SMBETS-OESHS algorithm undergoes and the performances are examined over various measures. The simulation outcome stated that the enhanced solution of the SMBETS-OESHS system over the recent techniques.

Due to the large demand for big data storage capacity, the storage intensity index is not calculated in the current big data cloud storage process, resulting in a high storage space usage. This paper proposes a big data efficient hybrid cloud storage model and algorithm under 5G network. The model is based on the 5G network performance framework and consists of three parts: users, private cloud and public cloud storage service providers. The purpose of efficient hybrid cloud storage of big data is achieved by using consistent hash algorithm. The simulation results show that the above algorithm occupies less storage memory space, the device load variance is smaller, the overall system load is more stable and balanced, and the average response is fast, which provides a favorable basis for the efficient hybrid cloud storage algorithm of big data.

Lung disease is considerable deprivation from health standpoint. These include chronic obstructive pulmonary illnesses, asthma, lung fibrosis, lung parenchyma illnesses, and tuberculosis among others. It is highly critical in the early phase of lung illnesses when they are the most treatable. Many of these were made for the purpose of applying machine learning and image processing. Many types of DL methods including CNN, VNN, VGG networks, capsule networks are used during lung illness prediction process. Following the release of the book on Pandemic Covid-19, many projects have been carried out at international level intending to study the feasibility of such work for prediction of future events. Pneumonia is a lung infection that starts earlier in the disease course and is closely associated with the virus (pneumonia condition), which was responsible for considerable chest infection in some covid-positive individuals. While doctors are no strangers to lung diseases and their complicated nature, many will find it difficult in some of them to make distinctions between common pneumonia and the Covid-19. X-ray imaging of the chest provides the highest degree of accuracy in suffem lung diseases. In this work, a novel approach for the calculation of lung illnesses such as pneumonia and COVID-19 is proposed. The data source for this method is Chest X-ray pictures taken from patients. The system includes characteristics such as the extraction of features, the prediction of illnesses, and the precise and adaptive evaluation of ROI, the collecting of datasets, and the enhancement of image quality. In future, this research can be extended with IOT devices for the recognition of COVID-19 and pneumonia.

Fatty liver disease, encompassing conditions like NAFLD (Non-Alcoholic Fatty Liver Disease) and NASH (Non-Alcoholic Steatohepatitis), is a significant global health issue linked to metabolic syndrome and increasing incidences of liver-related complications. Accurate and early detection of fatty liver illness is critical for effective intervention and management. This paper proposes a novel method for the prediction and arrangement of fatty liver disease using Probabilistic Neural Networks (PNNs), leveraging advanced machine learning techniques to enhance diagnostic accuracy and reliability. We developed a PNN-based model to classify liver conditions from a dataset comprising clinical and imaging features, including liver fat content, texture metrics, and demographic information. The PNN was chosen for its capability to handle complex, high-dimensional data and provide probabilistic outputs, which are crucial for assessing the likelihood of different disease stages and improving interpretability. The proposed methodology includes preprocessing steps to normalize and augment the data, followed by feature extraction using advanced techniques to capture relevant patterns. The PNN architecture was designed with multiple layers to process features and deliver class probabilities. The method's concert was estimated utilizing average system of measurement such as accuracy, precision, recall, and F1-score, demonstrating its efficacy in distinguishing between different stages of fatty liver disease. Experimental results indicate that the PNN model achieves high classification accuracy and outperforms traditional machine learning methods in detecting fatty liver illness. This study highlights the potential of PNNs in enhancing diagnostic processes and providing a robust tool for clinicians. Future work will concentrate on expanding the dataset, refining the model, and integrating it into clinical workflows to support better patient outcomes in liver disease management

Deep learning offers practical answers for neural network models when applied to cloud registering security. Via robotization Distinguish dangers, decrease manual checking, and further develop in general security adequacy. Deep learning network models assume a pivotal part in security errands like interruption discovery, malware identification, anomaly recognition, and log examination. requires Deep Learning mix in cloud security cautiously assesses existing frameworks, characterizes goals, chooses dataset with arrangement, model tuning and last changes for consistence. Moreover, applying deep learning methods in cloud security requires thought of variables, for example, computational assets, information assortment, arrangement costs, model turn of events, mix endeavors, and continuous observing and support. This study proposes an artificial neural network (ANN) model portrayal in the cloud to track down cloud security parts and recreate security techniques and researches the essential moves toward coordinate these models in the cloud. Regarding that the adequacy of the ANN scheme relies upon cloud parameters like the nature of the preparation information and the network architecture Also, weight change calculations. The review emploies a dataset from Kaggle.com to approve the recreation and blueprints the means Partake in preparing and assessment of the ANN structure.