Daily Archives: May 23, 2026

Authors: Mohammed Munyim Hussain V, Poorvaj K P, Prashanth S R, Preetham M, Mr P Prasanna

Abstract: Agriculture remains a cornerstone of economic activity across developing nations, yet smallholder farmers routinely face yield gaps caused by uninformed decisions on crop selection, soil nutrition, and disease management. This paper presents AGRI HUB, a web-based Crop and Soil Management System that unifies several machine-learning and deep-learning services behind a single Flask-driven interface. Four core modules are delivered: (i) smart crop recommendation using a Random Forest classifier trained on seven agro-climatic parameters, achieving 99.55% accuracy across 22 crop classes; (ii) soil nutrient analysis and fertilizer recommendation through NPK deficit computation against crop-specific thresholds; (iii) plant disease detection using a ResNet-9 convolutional neural network capable of classifying 38 disease categories from leaf photographs; and (iv) real-time, weather-driven activity planning by consuming OpenWeatherMap API data to generate seven-day farming calendars. An AI chatbot powered by the Google Gemini large language model supplements the analytical modules with conversational agronomic guidance. A crop profitability comparison dashboard rounds out the system, enabling evidence-based economic decisions. Experimental evaluation confirms that the integrated platform consistently outperforms single-module alternatives in both accuracy and decision breadth, offering a scalable, cost-effective tool for precision agriculture.

Published by: vikaspatanker

Authors: Parag Hossain

Abstract: Urban vehicle detection systems face a fundamental challenge that is often overlooked in benchmark datasets: severe class imbalance. In real-world traffic scenes, common vehicles such as cars appear thousands of times more frequently than critical but rare categories including ambulances, e-bikes, and motorcycles. This imbalance causes standard detectors to become biased toward majority classes, leading to unacceptable failure rates for minority class detection in safety-critical applications. In this paper, we propose a novel Class-Balanced Knowledge Distillation (CBKD) framework specifically designed to address this challenge on the challenging CAVI-14 dataset, which contains fourteen urban vehicle categories with up to fifteen-fold class imbalance. Our method integrates three key components: class-balanced sampling to ensure equal exposure to all classes during training, focal loss with class-specific weights to down-weight easy majority examples, and knowledge distillation from a teacher model pretrained on a synthetically balanced dataset. Extensive experiments demonstrate that CBKD achieves perfect mean average precision at 0.50 intersection-over-union threshold (mAP50) of 1.000 and near-perfect mAP50-95 of 1.000 after one thousand training epochs. Per-class F1 scores consistently exceed 0.97 across all fourteen categories, including the rarest classes. Qualitative results on validation images show accurate detection even under heavy occlusion and challenging lighting conditions. Our approach establishes a new state-of-the-art on the CAVI-14 dataset and provides a practical, reproducible solution for imbalanced object detection in intelligent transportation systems.

DOI: https://doi.org/10.5281/zenodo.20351972

Published by: vikaspatanker

Authors: Natalie Brooks, Grace Mitchell, Charlotte Evans, Amelia Foster, Naveen Kumar

Abstract: The rapid adoption of cloud-native architectures has increased the demand for high-performance Java applications capable of delivering scalability, reliability, and operational efficiency across distributed enterprise environments. This research paper explores intelligent JVM tuning and cloud scaling strategies designed to optimize the performance of Java-based cloud applications operating in modern hybrid and multi-cloud infrastructures. The study examines critical performance optimization techniques including garbage collection tuning, heap memory optimization, thread management, JVM parameter configuration, container-aware resource allocation, and real-time application monitoring. Additionally, the paper investigates the role of cloud orchestration platforms, Kubernetes-based auto-scaling, AI-driven observability systems, and predictive resource management frameworks in enhancing application responsiveness and infrastructure utilization. Intelligent automation mechanisms integrated with JVM performance analytics enable dynamic workload balancing, anomaly detection, and proactive remediation of performance bottlenecks. The research further analyzes the impact of microservices architectures, distributed caching systems, and continuous deployment pipelines on improving scalability and operational agility. Security, governance, and cost optimization considerations associated with enterprise-scale Java cloud deployments are also discussed. The findings demonstrate that intelligent JVM tuning combined with adaptive cloud scaling significantly improves application throughput, reduces latency, enhances fault tolerance, and minimizes operational overhead in high-volume enterprise computing environments. This research provides a comprehensive framework for organizations seeking to modernize Java application infrastructures while maintaining performance stability, business continuity, and long-term cloud operational efficiency.

DOI: http://doi.org/10.5281/zenodo.20351776

Published by: vikaspatanker

Authors: Michael Anderson, Matthew Collins, Daniel Foster, Christopher Hall, Naveen Kumar

Abstract: Modernizing legacy financial systems has become a strategic priority for enterprises seeking to improve operational agility, scalability, security, and digital service delivery in rapidly evolving financial ecosystems. Traditional financial platforms built on monolithic architectures and outdated technologies often suffer from high maintenance costs, limited interoperability, performance inefficiencies, and reduced adaptability to modern cloud-native environments. This research paper explores enterprise-scale re-engineering approaches for transforming legacy financial systems through Java-centric software paradigms integrated with intelligent cloud automation frameworks. The study examines the role of Java-based microservices architectures, containerization, API-driven integration, Infrastructure as Code (IaC), DevOps practices, and AI-powered cloud orchestration in enabling scalable and resilient modernization strategies. Furthermore, the paper analyzes how intelligent automation technologies, including machine learning, predictive analytics, automated deployment pipelines, and autonomous monitoring systems, enhance system reliability, operational efficiency, and infrastructure optimization across hybrid and multi-cloud financial environments. The proposed framework emphasizes secure migration methodologies, continuous compliance validation, self-healing operational capabilities, and cloud-native application modernization for mission-critical financial services. Additionally, the research discusses implementation challenges such as legacy system complexity, regulatory compliance, cybersecurity risks, data migration constraints, and organizational transformation requirements. The study concludes that the integration of Java-centric re-engineering methodologies with intelligent cloud automation frameworks provides a robust foundation for achieving sustainable enterprise modernization, accelerated digital transformation, improved customer experience, and long-term technological adaptability within modern financial institutions.

DOI: http://doi.org/10.5281/zenodo.20351718

Published by: vikaspatanker

Authors: B. M. Promod Kumar, Bhavana N. S., C. Chinmayi, Deepthi C. Shekar, Deenadayal B. K.

Abstract: Radiology reports generated from imaging modalities such as X-rays, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and ultrasound scans are critical clinical documents. However, these reports are authored in complex medical terminology intended for radiologists and specialist physicians, rendering them largely inaccessible to patients and non-medical users. This communication gap results in confusion, anxiety, and increased dependency on healthcare professionals for basic explanations. This paper presents MedLens, an AI- powered radiology report simplification system that bridges this gap by leveraging Natural Language Processing (NLP) and Retrieval-Augmented Generation (RAG). The system extracts text from uploaded PDF reports using PyMuPDF, processes clinical content using Google Gemini AI models, and generates accurate, context-aware patient-friendly summaries. It further classifies the urgency of findings into levels (Low, Moderate, High, Critical), and integrates multilingual translation, text-to-speech functionality, and an AI-powered contextual chatbot. The platform is deployed using FastAPI on the backend and React.js with Tailwind CSS on the frontend. Experimental results demonstrate that MedLens successfully simplifies complex medical terminology, detects critical conditions, provides multilingual support, and enables interactive report-based queries, thereby empowering patients with better health awareness and facilitating informed discussions with healthcare providers.

DOI: https://doi.org/10.5281/zenodo.20351715

Published by: vikaspatanker

Authors: Alexander Whitmore, Benjamin Clarke, Daniel Harrington, Ethan Montgomery, Naveen Kumar

Abstract: Autonomous Infrastructure Management using LLM-augmented platform engineering frameworks represents a transformative approach to modern cloud operations, combining large language models (LLMs), artificial intelligence, and platform engineering principles to automate infrastructure provisioning, monitoring, optimization, security enforcement, and lifecycle management across hybrid and multi-cloud environments. This research paper explores how LLM-driven automation frameworks enhance Infrastructure as Code (IaC), intelligent orchestration, self-healing systems, predictive analytics, and policy-driven governance to reduce operational complexity and improve infrastructure reliability. The study highlights the integration of natural language processing, machine learning-based anomaly detection, and autonomous decision-making mechanisms that enable adaptive infrastructure management with minimal human intervention. Furthermore, the paper examines the role of AI-powered observability, automated incident response, resource optimization, and compliance validation in accelerating DevOps and AIOps workflows while improving scalability, cost efficiency, and cybersecurity resilience. The proposed framework demonstrates how LLM-augmented platform engineering can streamline enterprise cloud operations through intelligent automation, contextual infrastructure recommendations, and continuous optimization strategies. Finally, the research discusses implementation challenges, ethical considerations, governance requirements, and future advancements in autonomous infrastructure ecosystems, emphasizing the growing significance of generative AI in next-generation cloud-native platform engineering and enterprise infrastructure transformation.

DOI: http://doi.org/10.5281/zenodo.20351661

Published by: vikaspatanker

Authors: Dr. Victoria S. Turner, Dr. Isabella N. Hughes, Dr. Christopher J. Walker, Prof. Daniel T. Harrison, Naveen Kumar

Abstract: Machine learning-driven infrastructure blueprinting and cloud architecture optimization represent a transformative approach to modern enterprise computing environments by integrating intelligent automation, predictive analytics, and adaptive resource management into cloud infrastructure design and deployment processes. Traditional infrastructure planning methods often require extensive manual intervention, static configuration models, and continuous monitoring efforts, which can lead to inefficiencies, increased operational costs, and scalability limitations in dynamic cloud ecosystems. This research explores the application of machine learning techniques in automating infrastructure blueprint generation, workload prediction, resource allocation, performance optimization, and fault detection across multi-cloud and hybrid cloud environments. By leveraging supervised learning, reinforcement learning, and deep neural networks, intelligent systems can analyze historical operational data, identify optimal architectural patterns, and generate scalable infrastructure configurations that align with business requirements, security policies, and compliance standards. The study further examines how AI-driven optimization improves cloud elasticity, reduces energy consumption, enhances infrastructure reliability, and accelerates Infrastructure as Code (IaC) deployment workflows through automated decision-making and self-healing capabilities. Additionally, the research highlights the integration of predictive analytics for proactive capacity planning, anomaly detection, and cost-aware cloud orchestration to improve operational resilience and service availability. The findings demonstrate that machine learning-enabled cloud architecture optimization significantly enhances deployment efficiency, reduces human error, strengthens infrastructure governance, and supports intelligent digital transformation initiatives in modern enterprises.

DOI: http://doi.org/10.5281/zenodo.20351605

Published by: vikaspatanker

Authors: Shital Vishwanath Ban, Shankar Sanjay Rathod, Prerana Prakash Malgave, Mrs. M.R, Raste

Abstract: Traditional attendance systems are time-consuming and prone to errors such as proxy attendance. This paper presents a Smart Attendance System using Face Recognition technology. The system automatically detects and recognizes faces to mark attendance. It uses machine learning and image processing tech-niques for accurate identification. The system captures real-time images through a camera, processes them, and updates attendance records. It reduces manual effort and improves accuracy. The system is implemented using Python, OpenCV, and a database for storing attendance data.

Published by: vikaspatanker

Authors: Ethan Cole Harrison, Prof. Daniel Reeves Walker, Prof. Emily Carter Hayes, Dr. Christopher Liam Foster, Naveen Kumar

Abstract: Infrastructure as Code (IaC) has emerged as a foundational paradigm for automating cloud infrastructure provisioning, configuration management, and deployment orchestration across modern enterprise environments. However, the growing complexity of multi-cloud architectures, dynamic scaling requirements, and heterogeneous deployment policies has increased the difficulty of maintaining reliable and secure declarative infrastructure templates. This research explores the integration of Generative Artificial Intelligence and neural modeling techniques into Infrastructure as Code workflows to enable intelligent, adaptive, and automated cloud infrastructure engineering. The proposed framework leverages large language models, transformer-based neural architectures, and AI-assisted configuration synthesis to generate, validate, optimize, and remediate declarative infrastructure definitions across cloud platforms. The study investigates how generative models can enhance infrastructure provisioning accuracy, reduce manual scripting complexity, improve deployment consistency, and accelerate DevOps and platform engineering operations. Furthermore, the research examines AI-driven policy validation, anomaly detection, infrastructure drift correction, security compliance automation, and predictive resource optimization within declarative cloud ecosystems. Experimental analysis demonstrates that neural-assisted IaC generation significantly improves deployment efficiency, operational scalability, infrastructure resilience, and automation intelligence while minimizing configuration errors and deployment failures. The findings highlight the transformative potential of generative AI in enabling autonomous cloud operations, intelligent infrastructure orchestration, and next-generation cloud-native automation frameworks for enterprise-scale digital transformation initiatives.

DOI: http://doi.org/10.5281/zenodo.20351381

Published by: vikaspatanker