Proceedings of the 7th International Conference on Communications and Cyber Physical Engineering

Table of Contents

1. Towards Sustainable Agriculture: Leveraging Advanced Technologies for Precision Crop Management and Optimization

   N. Haripriya, N. Venkatesh, S. Aravind Chary, P. Vineesha, K. Revathi

   Abstract

   Smart agriculture systems are a transformative solution for modern agriculture, addressing challenges like climate change and resource limitations. These systems gather data on environmental variables including soil moisture levels and weather conditions in real time using sensors, drones, and self-governing machinery—IoT gadgets. Predictive analytics is made possible by feeding this data into machine learning algorithms, which may detect possible dangers like disease outbreaks and insect infestations. By leveraging precision farming techniques, the smart agriculture system offers targeted and efficient resource allocation, optimizing water, fertilizers, and pesticide use. Continuous data analysis and machine learning model improvements enable informed decision-making for farmers. The integration of smart agriculture systems empowers farmers with real-time insights, enabling proactive measures to enhance crop yield and quality. The reduction in resource wastage contributes to a more sustainable agricultural ecosystem, reducing the industry's overall carbon footprint. These innovative technologies offer a promising pathway towards a resilient and efficient agricultural landscape, addressing pressing challenges related to food security and environmental preservation.

2. Asymmetric Reduced Switch Multi Level Inverter Topology for Renewable Energy Applications

   Anusha Oruganti, Gajula Ujwala

   Abstract

   In recent years as increasing the demand for electricity, the existing energy sources may not sufficient to satisfy the electrical demands, and responsible for polluting the environment. In order to meet the electricity demands, and maintain environment with pollution free. So people are focusing on renewable energy sources. For grid integration and AC load applications high voltage is needed as per the demand levels. Actually to increase the voltage levels high gain DC-DC or DC-AC converters and transformers are required. So to clear these problems, this paper introduces a reduced switch multi level inverter, it works on the asymmetric voltage sources and less no of switches for the application of renewable energy systems. A level shifted pulse-width modulation (LSPWM) scheme will be used in the proposed topology. The performance of this inverter can be validated through MATLAB/Simulink-based simulation results.

3. PromptAR: Image to Augmented Reality Using Unity Engine

   Venkanna Bhukya, P. Punitha, Shashank Tambe, P. Ashritha Reddy, K. Manish Pathri

   Abstract

   This paper aims to redefine the augmented reality landscape by seamlessly integrating user-friendly prompts with state-of-the-art technology. Through a user-centric approach, individuals can effortlessly craft personalized augmented reality objects, marking a significant leap forward in interactive, user-driven experiences. Prompts can be either scanned from images or given directly to the model. Use cases are endless as this can be used in educating students effectively with emphasis on visualizing the concepts. The platform's emphasis on personalized AR object creation empowers users to tailor their augmented reality content, offering a unique and immersive experience. With its application spanning various sectors, including education, design, healthcare, gaming, and manufacturing. The interactivity and engagement facilitated by this innovative approach not only redefine the possibilities of AR but also contribute to a broader evolution in how we interact with digital content. The integration of user-friendly prompts with state-of-the-art technology in augmented reality redefines the possibilities of how we interact with digital content. It enables individuals across various sectors to create personalized and immersive experiences, revolutionizing education, design, healthcare, gaming, and manufacturing. As this technology continues to evolve, the potential for innovation and transformative experiences becomes limitless.

4. Digital Transformation Chronicles: Sentiment Analysis on News Using Advanced Language Models

   P. Prakash, V. Sakthivel, L. Madhavan, T. Bharath, Harshita Rajinikanth, Poushikkumar Sivakumar

   Abstract

   The advent of the Internet has led to profound social changes, altering decision-making processes and social dynamics. However, this digital transformation comes with challenges, including the circulation of bad news that creates hopelessness. In response, sentiment analysis, a facet of natural language processing, has emerged as an influential tool for classifying media into positive, negative, or neutral categories. Our research highlights the contribution of sentiment analysis to classifying news content and reviews methods and models. We introduce a methodology encompassing data collection, preprocessing, and the use of models including BERT, XLM-RoBERTa, mBERT, DistilBERT, and XLNet. Using a dataset of 3000 labeled news stories, we train and evaluate sentiment analysis models. Data preprocessing involves steps such as tokenization, encoding, label encoding, and creating efficient data loaders using PyTorch. Metrics like precision, accuracy, recall, and F1 score give a well-rounded evaluation of the model's performance. In summary, this research highlights the significance of sentiment analysis in tackling negative content and provides valuable insights for making informed decisions in today's digital landscape.

5. Progressing Plant Leaf Disease Detection with CNN: Addressing Environmental Obstacles via Adaptive Normalization and Active Learning Strategies

   S. Anuradha, Nagella NagaPrasanna, Mitikiri Bhavani, Ravilla Sowmya, SajjanaGandla Harika, T. Brahmananda Reddy

   Abstract

   Plants play a crucial role in ecosystems as primary producers and are essential for the well-being of humanity, fulfilling diverse needs including sustenance, medicine, and the provision of raw materials for various industries. The project delineates a notable initiative aimed at addressing the persistent issue of plant diseases among smallholder farmers through advanced technology. Utilizing the widespread prevalence of smartphones and advancements in computer vision, the research centers on evaluating the effectiveness of Convolutional Neural Networks (CNNs), specifically focusing on the ResNet34 model, MobileNetV2 in swiftly and accurately detecting crop diseases. The CNN model, crafted and fine-tuned with a dataset encompassing 8,685 leaf images acquired under controlled conditions, attains commendable validation outcomes, boasting an accuracy rate of 97.2% and an F1 score surpassing 96.5%. The implementation of this model as a web application facilitates accessibility, empowering farmers to employ the technology for discerning seven distinct plant diseases amid healthy leaf tissue. This study underscores the practicality of incorporating CNNs in agricultural settings, representing a substantial advancement towards AI-driven solutions poised to fortify the resilience and productivity of smallholder farmers.

6. AutoInt: Automatic Feature Interaction Based Predictive Model for Kidney Risk Progression After Kidney Tumor Nephrectomy Using Clinical Data

   P. Suman Prakash, P. Kiran Rao, Venkata Pavani

   Abstract

   The evolution of kidney risk after kidney tumor nephrectomy is an important part of postoperative management. We provide AutoInt, an automatic feature interaction predictive model intended to evaluate the risk profiles related to nephrectomy in this work. Through self-attention techniques, AutoInt integrates complicated feature interactions using a broad set of clinical data, improving clinical decision-making and patient care in this particular setting. Our findings show that AutoInt outperforms baseline machine learning systems, achieving an amazing accuracy rate of 96%. The model offers unparalleled interpretability and predictive performance due to its automatic capturing of feature interactions that include both category and numerical features.

   AutoInt's potential goes beyond data analysis; by providing individualized medical therapies for patients who have had nephrectomy, it could transform the area of nephrology. Healthcare professionals may make quicker and more informed judgments by expediting the diagnostic process, which will ultimately lessen the postoperative burden for these patients. AutoInt has the potential to greatly impact kidney tumor patients’ postoperative care and therapy, resulting in better outcomes and a greater standard of living. AutoInt is a ray of hope for people traversing the complicated terrain of renal risk progression after nephrectomy, with its astounding 96% accuracy.

7. Computerized Cognitive Retraining Approach Integrating Personalized Activities and Machine Learning for Precision and Ethical Considerations

   S. Anuradha, Kanapuram Archana, Balusa Vasudha, Viratapu Tejaswini, Elluru Sai Rutwika, A. Pradeep Kumar Yadav

   Abstract

   This review explores the empirical evidence on the effectiveness of computer-assisted cognitive retraining for children with Specific Learning Disabilities (SLD). These children generally have average or above-average intelligence but face considerable challenges with language processing. Cognitive retraining is a therapeutic method aimed at improving underdeveloped cognitive functions through repeated practice. Grounded in the theory of brain plasticity, the training aims to enhance cognitive skills through carefully designed exercises and tasks. It can be administered manually or through computer-based systems, with the latter being particularly engaging, innovative, multisensory, and motivating for children. The purpose of this review is to compile significant research in the area and evaluate the effectiveness of these interventions for SLD children. In countries like India, where the large population and shortage of trained therapists make it difficult to provide direct access to SLD children, manual interventions may not be a practical solution. Therefore, scalable approaches that can reach remote areas must be investigated. This review aims to establish a foundation for examining the potential of computer-assisted retraining as a viable solution for SLD intervention.

8. Vein Secure Robust and Reliable Finger Vein Authentication System

   Sathyanarayana Cheruku, Swetha Pesaru, S. Vijay Kumar, Keerthana Kothoju, Asritha Donuru

   Abstract

   Finger vein authentication is a reliable biometric method with advantages like resistance to forgery and environmental variations. This paper explores the use of Convolutional Neural Networks (CNNs) to improve the accuracy and robustness of finger vein authentication systems. CNNs can learn discriminative features from raw input data, improving the effectiveness of finger vein recognition in real-world scenarios. The review covers recent research on CNN-based approaches for finger vein authentication, including network architectures, feature extraction techniques, and training strategies. However, challenges such as large-scale training datasets and computational resources are discussed. Future research in CNN-based finger vein authentication should focus on exploring novel network architectures, improving generalization across diverse finger vein patterns, and addressing privacy and security concerns.

9. Virtual Canvas Gesture Based Art Creation System

   Adilakshmi Siripireddy, Kukutla Kavya, Kareddy Yugendhar Reddy, Piske Vansh Raj, Muppidishetty Sai Ashwith

   Abstract

   Amid the difficulties introduced by the COVID-19 pandemic, the prominence of online education has significantly increased, requiring creative methods to effectively involve students. This initiative aims to meet the demand for interactive and captivating virtual learning platforms through the introduction of an innovative paint application that utilizes Media Pipe and OpenCV technologies. By integrating face detection and object tracking features, the application establishes a clean virtual classroom environment suitable for online instruction. Through the incorporation of facial gestures, users are provided with an instinctive approach to Human-Computer Interaction (HCI), ultimately enhancing the educational experience. The main objective of this endeavor is to enhance human-computer interaction in virtual educational environments, thus promoting more stimulating and interactive online learning opportunities for students.

10. Detection and Classification of Pneumonia in Chest X-ray Images Using Deep Learning Techniques

    B. Naveen Kumar, U. Rajesh, S. Mahaboob Basha, C. Seshikanth, K. Shailendra Kumar

    Abstract

    In the intricate realm of modern medical diagnostics, the crucial goals revolve around the nuanced identification and categorization of pneumonia gleaned from chest X-ray images. Pneumonia, a respiratory infection, tends to proliferate in areas grappling with heightened pollution, unsanitary living conditions, and inadequate medical facilities. The imperative of improving treatment outcomes and bolstering survival rates places pneumonia diagnosis at the forefront, demanding swift and pinpoint precision. However, the visual interpretation of chest X- rays persists as a challenging and inherently subjective endeavor. This endeavor embarks on employing cutting- edge deep learning techniques to surmount extant quandaries surrounding pneumonia diagnosis. The strategy involves deploying deep transfer learning to circumvent the dearth of labeled datasets due to data availability constraints. With an unwavering emphasis on refining accuracy, a pioneering Convolutional Neural Network (CNN) model takes shape, seamlessly amalgamating CovXNet, RNN, VGG16, and an evolutionary upgrade beyond the prevailing ResNet 50. This research venture is poised to confront emerging challenges, including imbalances within datasets, the subjective nuances inherent in X-ray interpretations, and the optimization intricacies of transfer learning techniques. Through a judicious blend of critical analysis and methodological refinement, this scholarly pursuit aims not only to furnish a robust Comprehensive Diagnostic Infrastructure for Pneumonia (CDIA) but to usher in substantive enhancements in precision and applicability across diverse patient demographics and multifarious clinical settings, promising a transformative impact on the landscape of medical diagnostics.

11. Analysis and Prediction of Crime Hotspots Using Machine Learning with Stacked Generalization Approach

    S. Mahammad Rafi, B. Naveen Kumar, K. Balaji Reddy, J. Pavan Kumar, P. Datta Sai Prem Prathap Reddy, P. Harith

    Abstract

    The ensemble learning approach is a cooperative decision-making system that generates new examples by combining the predictions of learnt classifiers. The early analysis has demonstrated that the ensemble classifiers are far more reliable than any single part classifier, both conceptually and experimentally. Even with the presentation of multiple ensemble approaches, determining the right configuration for a given dataset remains a challenging process. The goal of crime prediction is to discourage criminal activity and lower the crime rate. This paper presents the assemble-stacking based crime prediction method (SBCPM), an authentic and efficient approach based on algorithms for determining the right crime predictions. Learning-based methods are applied to achieve domain-specific configurations that are compared with another machine learning model. The ensemble model that has the highest correlation coefficient and the lowest average and absolute errors sometimes performs better than the other models. On the test set, the suggested approach produced accurate classifications. It was discovered that the suggested method was helpful in forecasting potential crimes and imply that the ensemble model's prediction accuracy is greater than the individual classifier's.

12. Future Effectual Aspects of Surveillance at LoC: A Systematic Analysis of Smart Sensors, IoT and Drones

    Dasari Naga Vinod, R. Asha, G. Kavitha, N. Kapileswar, Judy Simon, Polasi Phani Kumar

    Abstract

    In the present ecology of surveillance at the Line of Control (LoC), the commitment of the Internet of Things (IoT) to drones has procured gigantic plausibility because of its sophisticated favorable circumstances in different areas. IoT clears an approach to relate control of everything in pretty much every field of society. Alternatively, drones are considered general research, and blending IoT with drones together exhibits gigantic potential development plausible. This review article features the ultimate research works that pay attention to employing the Internet of Things in drones. The work likewise emits numerous creative methodologies utilized in IoT and drones and their particular applications in different areas and combat the COVID-19 epidemic. This paper aims to benefit researchers as well as new aspirants in the region of IoT, and drones as well as initiate perception for novel multidisciplinary research.

13. Eye-Guided Cursor Control: Precision Through Vision

    Bethi Sai Teja, S. Srinivas, Nakka Nishitha, Poloju Yeshwanth, Mereddy Suryaprakash Reddy

    Abstract

    This study investigates an interesting way of computer interaction that uses the natural movement of the eyes to control the on-screen pointer. Traditional mouse and keyboard methods might be challenging at times, particularly for people with physical restrictions. The suggested eye-ball-based cursor movement system seeks to provide a more intuitive and accessible alternative. In this study, we look at the technology behind eye-tracking gadgets and how they fit into everyday computing. Understanding how the eyes move allows us to design a responsive system that lets users control the cursor with their gaze. The simplicity of this technology allows for a more inclusive computing experience for everyone. The study looks into the potential applications of eye-ball-based cursor movement, such as gaming, accessibility features, and improved user interfaces. Furthermore, we address concerns about privacy and user comfort, ensuring that the technology respects personal limits. Through this research, we hope to contribute to ongoing efforts to make technology more user-friendly and accessible, ultimately building a more inclusive digital world for individuals of all abilities.

14. Innovative Deep Learning Framework for Accurate Fault Diagnosis in Industrial Systems

    T. Harikrishna, G. Nandini, G. Peerambi, N. Manasa, B. Pallavi

    Abstract

    As industrial systems advance towards automation, the demand for accurate fault diagnosis methods grows. Deep learning (DL) faces challenges such as parameter volume and initialization instability. To overcome these hurdles, we proposed a DL framework integrating CNNs and TL. Signals are converted into 2-D images by Continuous wavelet transformation (CWT) improving representation. SVMs replace fully connected layers, enhancing efficiency. The method outperforms classical DL architectures. Validation includes convergence curves, testing reports and confusion matrices. It achieves highest accuracy in cross-domain diagnosis and fault detection. Across mechanical datasets, it sets a new standard for performance.

15. Smart and Effective Real-Time Management of Street Parking

    K. Kavitha, Y. Chandana, E. Amulya, P. Hima Varshini, K. Mehatab

    Abstract

    With the rapid growth of the automotive industry, passenger cars are becoming increasingly common in large cities, making it harder to find available parking spaces. Parking lots represent a heavily used service, with significant investments made each year. Managing these parking facilities can be costly and complex, particularly in large-scale areas such as airports or shopping centers. Traffic congestion in parking areas is a growing issue, as the number of vehicles rises much faster than the availability of parking spaces. The proposed model identifies all available parking slots in a given area and analyzes data to determine whether each slot is vacant or occupied, providing real-time information on available spaces. Our system is designed to detect the parking slot availability by using CNN, Random Forest and SVM. These methods were evaluated using precision, recall and accuracy. The results obtained were tested, SVM accuracy is 98% which is promising in finding the vacant slots of the parking area.

16. Machine Learning for Early Detection of Dementia: A Predictive Modeling Approach

    G. Fayaz Hussian, S. Mounika, M. Amrutha, R. Swathika Rani, P. Divya

    Abstract

    Dementia is a degenerative neurological condition that worsens with time and is characterized by a deterioration in cognitive abilities and difficulties in carrying out everyday activities. Timely identification of dementia is essential for prompt management and improved patient outcomes. Machine learning algorithms have shown potential in detecting patterns and signs linked to the development of dementia in recent times. This research introduces an innovative method for identifying dementia at an early stage by using machine learning algorithms on a range of datasets, such as medical records, neuroimaging scans, cognitive evaluations, and demographic information.The suggested technique encompasses the processes of data collection, preprocessing, feature selection, and the building of a machine learning model. Data preparation strategies address the issue of missing data, standardize features, and eliminate noise in order to provide machine learning models with high-quality input. The dataset is analyzed to identify and extract pertinent aspects that indicate the course of dementia. Multiple machine learning methods, including logistic regression, support vector machines, random forests, and deep learning architectures, are used to train and optimize the classification of people as either having dementia or being healthy.

17. Identifying Deforestation Using AI Enabled Satellite Image Processing

    G. Fayaz Hussain, G. Asritha, G. Sowmya, C. Charitha, D. Susmitha

    Abstract

    Detecting deforestation through AI-assisted satellite image analysis involves utilizing AI techniques to accurately identify deforested areas by analyzing satellite images, distinguishing them from natural vegetation. This process allows for timely detection of large-scale forest cover changes. Precision is crucial, especially in challenging conditions like cloud cover, while scalability and real-time processing pose additional obstacles. The aim is to efficiently address these challenges to promptly recognize deforestation patterns across vast areas. This approach harnesses AI-powered satellite image analysis to enhance detection accuracy and speed amidst significant environmental shifts. The proposed solution employs a statistical model integrated with an adaptive AI system, combining advanced algorithms like CNNs with techniques such as spatial autocorrelation analysis to enhance accuracy and address scalability issues. Statistical methods handle data preprocessing, feature extraction, and dimensionality reduction, facilitating efficient recognition of forest cover changes on a large scale. This system surpasses current methods by offering superior accuracy and precision in identifying deforestation through AI-powered satellite image processing, while also reducing time complexity. The AI-based model demonstrates promising results, achieving 87% accuracy and 85% precision, with impressive computational efficiency, suggesting potential for real-time analysis in extensive deforestation monitoring efforts. This underscores the effectiveness of AI-driven satellite image processing in monitoring deforestation on a large scale.

18. Credit Risk Analysis Using Ensemble Machine Learning Models

    Sunil Bhutada, Mohd Aiman Saleem, Munaga Nandini, B. N. U. Chandrika

    Abstract

    The precise evaluation of credit risk continues to be a crucial component of prudent decision-making and risk management in the banking and lending industries in the ever-changing financial landscape of today. While traditional methods of credit risk analysis, which frequently depend on isolated modeling methodologies, have proven effective, they might not be able to fully represent the complexity of today's financial settings. The need for methods that can provide increased predictive power and adaptability grows as markets change and become more complicated. To address this problem, ensemble approaches have surfaced as a strong contender, offering a framework that combines the predictive power of several models into a coherent whole. This study uses a range of machine learning algorithms, including XGBoost, CatBoost, Decision Tree, Logistic Regression, KNN, and Random Forest to explore the potential in the field of credit risk analysis. By leveraging the unique properties of many algorithms within an ensemble framework, the objective is to improve forecast accuracy while also strengthening the robustness and adaptability of credit risk assessment approaches. This introduction discusses how ensemble approaches can revolutionize credit risk analysis and establish the groundwork for a full discussion of them. It also offers insights into practical implementation considerations and empirical validations.