International Institute of Information Technology

Abstract- The rapid expansion of smart residential environments requires integrated systems that combine security, energy optimization, and real-time monitoring within a unified framework. However, many existing solutions operate as isolated modules without embedded artificial intelligence for identity verification and dynamic energy management. This paper presents an AI-enabled smart home system that integrates vision-based authentication, occupancy-aware automation, and real-time energy analytics using an edge-computing architecture. The system is implemented on a Raspberry Pi 4 Model B and employs Haar Cascade-based face detection for owner authentication. Unauthorized access triggers automated email alerts to enhance security. Real-time electrical parameters, including voltage, current, and power consumption, are measured using a PZEM-004T, enabling continuous monitoring and automated monthly billing estimation. An occupancy-driven control mechanism using PIR sensing ensures automatic appliance shutdown during idle conditions, reducing energy wastage. A web-based dashboard provides live system monitoring. Experimental results demonstrate reliable intrusion detection and improved energy efficiency. The proposed framework offers a compact, scalable, and cost-effective solution for secure and sustainable smart home environments. Keywords: Smart Home System, Artificial Intelligence, Face Detection, Energy Monitoring, Occupancy-Based Automation, Edge Computing, Intrusion Detection, Energy Optimization, Embedded Systems.

Plant diseases threaten food security, especially in pulse crops essential for sustainable agriculture. This study explores Convolutional Neural Networks (CNNs) for rapid detection of major diseases such as Ascochyta blight, Fusarium wilt, and phytoplasma infections, which cause severe losses in countries like Ethiopia, India, and Taiwan. Using advanced imaging, we propose a web-based diagnostic tool to empower farmers with timely, accurate disease identification, enabling effective management decisions. This approach offers not just technological advancement but also economic sustainability for smallholder farmers by improving crop health and yields through early intervention and precision agriculture practices.

The print media is ever getting pressurized to produce output of high quality, customized output, shorter turn-around time, low operation cost and enhanced sustainability. The customary procedures of print media management, which are manualized in coordinating, fixed scheduling as well as fragmented data disclosures are progressively insufficient to the requirements. The given paper explores the implementation of intelligent workflow automation in the management of print media using the combination of artificial intelligence, machine learning, Internet of Things (IoT), and digital twin technologies. A perception-cognition-action-based paradigm is suggested to be the structured AI-driven workflow architecture to allow the adaptive data-driven orchestration between pre-press, printing, post-press, and distribution stages. Predictive quality control, dynamic scheduling, anomaly detection, and resource optimization are performed with the help of machine learning and decision-support models, whereas real-time physical-virtual synchronization and scenario-based planning is performed with the help of digital twins. The quantitative assessment indicates that there are great improvements in turnaround time, schedule compliance, machine use, quality rework, downtime, material waste, and energy usage. The findings reveal that intelligent workflow automation has multidimensional benefits of efficient operations, cost-saving, and sustainability, and maintains human controls by providing supervisory decision-support interfaces. The research paper finds that intelligent workflow automation is a strategic asset in the production workflow of contemporary print media organizations that aim to have resilient, efficient, and sustainable production systems.

Folk music is a cultural reminiscence, social ideals, and historical records of people, but numerous traditions are struggling with the issue of continuity in the environment of globalization and shifting musical realities. This article discusses how modern jazz improvisation has been used as a creative and culturally sensitive tool to maintain and recontextualize folk music traditions. Instead of seeing preservation as a conservation that is unchanging, the study conceptualizes heritage as a living process that continues to survive by changing. Using ethnomusicological viewpoints, jazz improvisation theory, and intercultural theory the study examines how rhythmic patterns, modal systems and performance practices of folk music can be incorporated in the modern jazz environment without the erosion of cultural identity. The paper shows that improvisation makes it possible to have a dynamic dialogue between tradition and modernity and the folk elements can keep the symbolic and emotional meaning they have and broaden their expressive opportunities that are achieved through harmonic reinterpretation, rhythmic adaptation, and group performance. The problematization of jazz improvisation as a cultural negotiation and sustainability points to the research as a means of heritage preservation through a living archive and pedagogical tool, as well as an ethical platform. The results provide a wider-ranging discourse on the field of musicology and performance studies by establishing preservation through transformation as a possible paradigm of maintaining traditional music in the context of the contemporary artistic practice.

The disappearance of modernist artworks through the historical conflict, material degradation and partiality of archival regimes poses a continuous problem in terms of art history and cultural heritage conservation. It is a foregone conclusion that such works are difficult to reconstruct, since modernist aesthetics is focused on abstraction, fragmentation, and conceptual purpose over representational devotion. In this paper, the author suggests a reconstruction model based on AI that views artistic reconstruction as a probabilistic inference problem, as opposed to a deterministic recovery problem. The system is already based on the framework of representation learning, conditional generative modeling, uncertainty-aware sampling, and human-in-the-loop validation to produce several possible discovery hypotheses based on the existing evidence and style samples. The hierarchical style modeling strategy gives reconstruction possibilities at artist and movements levels, whereas explicit uncertainty signs circumvent overconfidence in interpretation. The framework is tested by using representative case studies which are partially documented, partially fragmented and entirely lost modernist artworks. The quantitative metrics of consistency, the analysis of perceptual styles, and uncertainty analysis and the qualitative evaluation process performed by experts prove that the reconstruction fidelity is proportional to the strength of evidence and that the uncertainty is explicitly magnified in the case of sparse constraints. The findings indicate the significance of AI as a supportive analysis tool that can aid the judgment of curators and art-historians but does not sacrifice the interpretative responsibility.

Student drop out of the art education programs in the academic institutions is a major problem because the students tend to drop out of the programs due to a complex interplay of creative, behavioral, psychological, and the socio-economic factors and not as a result of their work performance. This research paper introduces a predictive dropout analysis model that suits the field of art education management, and which can be used to predict potential at-risk students at an early stage and effectively implement data-driven, time-based response. The framework combines institutional data that is heterogeneous, such as attendance data, studio and portfolio submissions, assessment data, and traces of use of digital tools with psycho-creative data, such as creativity indices, portfolio development rates, and qualitative feedback on critique. The advanced feature engineering methods are used to extract the measures of engagement-trajectories, skill-growth-slopes, and creative-consistency indicators of the longitudinal dynamics of learning specific to art-based programs. Several machine learning models, which include logistic regression, random forest, support vectors machines, artificial neural networks, and gradient boosting are trained and tested through a structured training-validation pipeline through hyperparameter optimization. The accuracy, area under the ROC curve, F1-score, and the precision-recall are the measures of model performance that are evaluated to guarantee that the model can be robust under class imbalance conditions. The experimental findings show that ensemble and non-linear models have a higher performance compared to the traditional baselines and show the predictive power of creative interaction and behavioral characteristics as well as academic indicators.