Pomegranate is one of the most well liked and nutritious fruits in the world. Pomegranates are now processed into a variety of food products and nutritional supplements due to the growing demand for them. Nowadays, pomegranate fruits are affected by diseases. In this paper, an Efficient Net classification is proposed to quickly identify diseases in pomegranate fruit. Firstly, histogram equalization is applied to the pomegranate fruit disease dataset to modify an image’s pixel values based on its intensity and by using laplacian filtering edges are detected to get high quality image. Next, the processed image is given to the K Means clustering for the separation of relevant regions. After that pomegranate fruit disease is featured using Scale Invariant Feature Transform (SIFT) for further analysis. Finally, Efficient Net framework is processed to enhance the classification of pomegranate fruit disease. Using python software the proposed framework is examined and the developed classifier results with accuracy of 95% in detecting disease when compared to other techniques.

Road maintenance is crucial for urban infrastructure, with potholes posing risks to vehicle safety and driver comfort. Traditional pothole detection methods are labor intensive, slow, and prone to errors. To address these challenges, we present a deep learning based pothole detection system leveraging the YOLOv8 architecture, known for its speed and accuracy. Developed in Python with a frontend comprising HTML, CSS, and JavaScript, the system is deployed via the Flask web framework. The detection algorithm, trained on a dataset of 780 images, achieves an accuracy of 71%. It offers three operational modes: image based detection, video processing, and real time detection using a webcam. The image based mode analyzes static images for potholes, while the video mode processes continuous footage for road monitoring. The real time detection mode integrates seamlessly with vehicle systems or roadside monitoring stations. Each mode utilizes the YOLOv8 model to detect potholes efficiently, enabling timely interventions for road maintenance. This project demonstrates the feasibility of deep learning in infrastructure monitoring, highlighting its potential to improve road safety and maintenance efficiency through accurate and scalable detection techniques. Road maintenance is a critical aspect of urban infrastructure management, with potholes posing significant hazards to vehicle safety and driver comfort. Traditional methods of pothole detection are often labor intensive, time consuming, and prone to human error. To address these challenges, we present a novel solution for road pothole detection leveraging deep learning techniques. This project employs the YOLOv8 architecture, a state of the art object detection model known for its high speed and accuracy. Our system is developed using Python, with a frontend composed of HTML, CSS, and JavaScript, and is deployed using the Flask web framework. The detection algorithm was trained on a dataset of 780 images, achieving an overall accuracy of 71%. The detection system is versatile, offering three modes of operation: image based detection, video based detection, and real time detection using a webcam. The image based mode allows users to upload static images, which are then analyzed for pothole presence. The video based mode processes video files, enabling continuous monitoring of road conditions. The webcam mode provides real time detection, making it suitable for integration into vehicle systems or roadside monitoring stations. Each mode leverages the YOLOv8 model to quickly and accurately identify potholes, providing valuable data for timely road maintenance interventions.

Existing traffic monitoring systems face significant limitations in ensuring road safety and enforcing traffic laws. Traditional systems often rely on manual intervention for identifying violations such as speeding or illegal parking, leading to inefficiencies and delays. Many existing solutions use outdated object detection methods that fail to provide real time, high accuracy results, especially in high density or low light traffic scenarios. Additionally, these systems lack integration for automated notifications and incident reporting, making it difficult for authorities to respond promptly. The proposed intelligent vehicle tracking system addresses these limitations by integrating advanced artificial intelligence and computer vision technologies. Vehicle detection is performed using the YOLOv8 model, renowned for its high accuracy and real time processing capabilities, ensuring precise identification of vehicles in various conditions. Speed monitoring leverages OpenCV to calculate vehicle speeds based on timestamps and predefined distances, while license plate recognition is achieved using EasyOCR, which extracts text from plates and links it to an owner database. The system includes an incident reporting module, allowing users to upload accident photos, with geolocation APIs automatically providing the device’s location for immediate response by law enforcement. Comparative testing revealed that the proposed system achieved a 95% success rate in vehicle detection and 92% in number plate recognition, surpassing the 85% accuracy of traditional methods. Furthermore, speed monitoring in the proposed system demonstrated consistent reliability with a ±3 km/h error margin, compared to higher error rates in existing solutions. The automated notification and incident reporting features significantly improve response times, addressing a critical gap in conventional systems. This system represents a significant step forward in leveraging technology for smarter traffic management, offering scalability for deployment in both urban and rural areas. Future enhancements include integration with national vehicle registries and predictive analytics for accident prone zones.

In the modern environment, it is quite difficult to detect Heart Disease (HD) through early stage symptoms. Heart disease is the cause of mortality if the diagnosis is delayed including death occurs to prevent these issues. This paper proposes an innovative approach for the detection of heart disease from HD dataset, incorporating advanced techniques. Initially, data cleaning and data normalization is applied to the HD to eliminate or to find the duplicate values.  Next, the images are segmented using K means clustering, allowing for the separation of relevant regions associated with HD. Features are then optimized from the segmented HD using the Dung Beetle Optimization (DBO) algorithm, capturing optimized values for further analysis. Finally, a Bidirectional Long Short Term Memory (BiLSTM) framework is proposed as a hybrid model to enhance the classification of HD images, leveraging both forward and backward temporal information to overcome challenges in disease analysis. The assessment of proposed work using python software reveals that the proposed framework with BiLSTM classifier ranks with improved accuracy of 91.56% when compared to the other techniques.

Tsunami detection in the coastal area is a major problem and a big impact on the environment, therefore early detection and training for tsunami need to be carried out to reduce the impact of casualties and losses incurred. Machine learning techniques, with their ability to identify patterns and anomalies, offer an approach for tsunami detection. This paper introduces a novel tsunami detection model combining the Chicken Swarm Optimization (CSO) Algorithm and Artificial Neural Network (ANN) classifier. Using the data pre processing technique such as data cleaning and data visualization the input dataset is processed and featured using speedup data transformation the raw data is converted in to certain data using feature engineering. This hybrid approach effectively analyses the detection of tsunami by leveraging the strengths of both methods. The model is trained and evaluated using the earthquake dataset, achieving a commendable accuracy of 92%. The software used here is python. The proposed model demonstrates superior results compared to existing techniques, offering an efficient solution for environment.

Credit card fraud has played a major issue for both cardholders and the issuing authorities as a notable financial challenge. To address this issue here involves modified Convolutional Neural Network (CNN) to identify fraudulent activities. The proposed methodology involves data pre processing, data visualization, validation and evaluation. Here the data is preprocessed by normalization to improve the quality of data performance and training stability of deep neural network. One hot encoding is used to convert categorical variables into binary format and is highly essential for data processing which in turn improve the prediction and classification accuracy of the method. Furthermore by using CNN with a modified Quantum Vortex Search Algorithm (MQVSA) is developed to increase the accuracy and effectiveness of the credit card fraud detection. The MQVSA improves CNN capacity of classification and to identify complex pattern indicative of fraud and model parameters. Thus the proposed method used in this work is performed in the python software and thus the classifier achieve the remarkable precision of 98% and the sensitivity value of 97% and effectively predicting the credit card fraud by evaluation and validation.

Sophisticated security measures are required to safeguard vital infrastructure and sensitive data due to the ever changing nature of cyber threats. Integrating deep learning techniques into cybersecurity is the subject of this study, which aims to improve threat detection, response, and prevention using novel methodologies. We compare the performance of different deep learning architectures, such as RNNs and CNNs, in detecting patterns that can indicate cyberattacks, using tools like malware detectors and intrusion detection systems (IDS). By utilizing massive datasets and real time analytics, these models outperform conventional methods in terms of speed and accuracy. Moreover, we offer real world examples of deep learning in action, including automated threat intelligence analysis and anomaly detection in network data. Strong training processes and model optimization are necessary, as the results show that cybersecurity frameworks must constantly learn and adapt. Also covered are issues with data privacy and the need for computational resources, two of the many obstacles to using deep learning solutions in cybersecurity. By expanding our knowledge of how deep learning can strengthen cybersecurity measures, this research ultimately leads to more resilient systems that can withstand emerging cyber threats.

The disease needs to be treated cautiously because of the complexity of brain cancer. Since each patient’s brain tumor is unique in terms of dimension, shape, and localization, it is challenging to identify them in medical imaging. Magnetic Resonance Imaging (MRI) process is used to diagnose the brain tumor anatomy results in inaccurate detection of the tumor and it is very time consuming. Therefore, many classification techniques from Machine Learning (ML) and Deep Learning (DL) are utilized to rapidly recognize the tumor. This paper proposes segmenting and detecting the brain cancer using Deep Convolutional Neural Network (DCNN). An image is first pre processed using a median filter to minimize noise in the image.  Global thresholding is a simple image segmentation technique that uses a single threshold value to separate an image into distinct regions. After that, the feature extraction method of Histogram of Oriented Gradients (HOG) is employed to detect brain tumors with higher accuracy and lower computational complexity than other methods. This performance measure is applicable to quantifying the performance of the segmented tumor region. Additionally, DCNN classification algorithm has achieved high performance and accuracy. For the purpose of identifying the performance of the model, this proposed project produced a confusion matrix. With a classification accuracy of 91.23% and a precision of 90.39%, the experiment’s results demonstrated that the method is more successful than the current methods.

The Automatic crack detection system is a significant technology to detect crack in concrete images. Identification of the crack Intensity and severity is the main objective of the detection system. Detection of images containing unwanted noises, blurriness and interference is considerably difficult hence image pre processing is done using the adaptive filter by which the disturbances is removed. Images with non uniform light, poor contrast or sudden brightness changes is rectified in the segmentation process using the fuzzy c algorithm that uses clustering technique. Techniques like Gray scale processing do not provide accurate detection therefore Gray level co event matrices (GLCM) technique which extracts high level features needed in order to perform classification of images and Multi support vector machine (SVM) is used because they can handle multiple continuous and categorical variables. The proposed technique provides better outcomes for crack detection and treatment. The PYTHON software is utilized.

Diabetic retinopathy is an eye condition which leads to the loss of eye sight in the person who already has diabetics. There are several stages in detecting the diabetic retinopathy which includes manual medical methods or detected by automatic methods using image processing. In these cases, neural networks are used in detecting the defects caused in the retina of the diabetics. These neural network process as same as human brain, in this methodology Artificial Intelligence (AI) is implemented. In this methodology, the image is pre processed to improve the image clarity by Gaussian filtering method. The pre processed image undergoes segmentation using the Fuzzy C means (FCM) clustering method. By involving these methods, the complexity in detecting the diabetic retinopathy is reduced. The Deep Convolutional Neural Network (DCNN) method is implemented in which data is compared to find the defect. The pooling layer samples the image to speed up computation, and the fully connected layer provides the final prediction. In the Convolutional Neural Network, the data to the compared is already processed and stored. This paper provides excellent specificity and sensitivity for classifying images as having or without diabetic retinopathy. This paper is implemented using Python Google Colab software.