Authors - Sahil Shelote, Ritesh Chaudhari, Payal Sirmokadam, Rupali Kamathe, Meghana Deshpande, VandanaHanchate, Sheetal Borde Abstract - Traditional traffic enforcement methods pose significant challenges to public safety in order to effectively detect and resolve violations. Using the ESP32-Cam module for video capturing, YOLOv3 for object detection, and OCR for license plate recognition, it offers an innovative approach to improving road safety and traffic management. ESP32-CAM module captures realtime videos of intersections. What sets this research work apart is the integration of YOLOv3, an advanced object detection model, to detect possible traffic violations such as helmet detection, rider detection. OCR technology allows extraction of license plate information, ensuring accurate identification of the vehicle involved in violation. Enabling the creation of Echallans and sending the registered vehicle owner an SMS with the payment gateway link when an Echallan is generated. This represents an important development in traffic management and safety, with promising results in terms of increased compliance, reduced accidents and general improvements in road safety. ESP32-CAM integrates YOLOV3 and OCR technologies to provide an efficient and technologybased solution to improve public safety on the road.
Authors - Rohini Hongal, Supriya K, Rajeshwari .M, Rahil Sanadi Abstract - Computer vision applications like object detection, picture matching, 3D reconstruction, and depth estimation in navigation rely on the synchronization of stereo frames. In stereo vision, two cameras separated by known distance are used to capture an image and analyze for differences in both images. To use stereo images in any application, synchronization between the corresponding frames must be ensured. This paper presents an approach to detect the synchronization between the stereo pair images. The synchronization information between the stereo frames can be achieved in two ways: one is by using the temporal data of the image pair and the other is by analyzing the spatial data in the images. This study uses the temporal data i.e. timestamps of the stereo images and validates results with the spatial data, to identify the stereo image pair as synchronous or asynchronous. The spatial algorithm is executed once the timestamp algorithm identifies a possible synchronization. In order to generate a template and extract spatial information from the left frame, this technique makes use of the Sobel filter. An appropriate correlation approach is then used to match the template to the right, right+1, and right-1 frames. If the chosen frame matches the correct frame, the frames are deemed to be synchronized. The frame with the highest correlation is chosen. On the other hand, the frames are considered asynchronous, if the frame with the highest correlation is either the right+1 or right- 1 frame. The suggested approach offers an accuracy of 90.33for static datasets and 96.67frame synchronization. The technique also provides information on the duration of asynchrony when frames are not synchronized. A variety of computer vision applications that depend on synchronized stereo frames might benefit greatly from the presented technique. It allows for more reliable object detection, picture matching, and 3D reconstruction by precisely detecting the synchronization state, which improves visual perception and comprehension in real-world circumstances.
Authors - Karuppasamy M, Jansi Rani M, Poorani K Abstract - Diabetes is the leading cause of mortality since its prevalence is higher globally. Since it contributes to various kinds of complications it leads to a high mortality rate. Early diagnosis and prediction of contributing features are found with the assistance of machine learning models. These models are instrumental in assisting healthcare sectors in prediction, diagnosis, prognosis, and disease prevention. If diseases are found at earlier stages, it would save many people’s lives. In that aspect, machine learning models are developed to find diseases at earlier stages. However, accuracy of the predictions at not much satisfied. This proposed work explores the techniques to predict diabetes at earlier stages. Several data mining approaches to XAI are discussed. The major features contributing to diabetes are also identified with the feature importance technique. This results in a greater way of understanding which feature contributes more to diabetic progression. The proposed model resulted in 94% accuracy with random forest which is also elaborated with Explainable AI (XAI).
Authors - Payal Khode, Shailesh Gahane, Arya Kapse, Pankajkumar Anawade, Deepak Sharma Abstract - An important subject that has always remained on top of the most important areas of concern universally is security as the world deals with dynamic change in technology. It is with this background that this paper explores the frailties that arise from the current technological gadgets such as mobile phones, Internet of Things (IoT) devices, and personal computers that are prone to a range of cyber threats. A comprehensive examination of the security threat is taken to show how application weaknesses and system susceptibilities and network-based threats allow the attacker to erode user confidentiality and data integrity. Moreover, this study compares traditional and modern assessment and protection mechanisms, including cryptography techniques, flow inspection tools, signals intelligence technologies, and hardware-based and artificial intelligence-based security measures with the intention of identifying the most effective paradigm for combatting these threats. That way, the present paper is relevant to the ongoing work in the field aiming at designing new countermeasures to improve the vulnerability of assorted present-day technologies to cyber threats.
Authors - N V Bharani Subramanya Kumar, C V Mahesh Reddy, CH. Samyana Reddy, Krishn Chand Kewat, Laxmi Narsimha Talluri, Shaik Mohammed, Rahil Sarfaraz, Sushama Rani Dutta Abstract - This work showcases an improvement over existing methods by developing a novel deep convolutional neural network (CNN) architecture for image classification specifically targeting the images in the CIFAR-10 dataset [4] which consists of 60,000 color images ( 32 x 32 pixels size) divided into 10 classes. So far, the model architecture incorporates a number of convolution and pooling layers which are then followed by the fully connected layers to better learn the complex structure existing within the input spatial configuration. The typical challenge of overfitting is addressed by employing various techniques such as data augmentation and dropout regularization strategy. Immediately from the experimental evidence, it is clear that the deep CNN performs superior to other traditional models in the case of image recognition classifying problems and therefore the model has proved to be robust in discerning the differences that exist in the categories in the images within the CIFAR-10 dataset.
Authors - Shubham Kadam, Chhitij Raj, Pankajkumar Anawade, Deepak Sharma, Utkarsha Wanjari, Vijendra Sahu, Anurag Luharia Abstract - This paper examines the modern role of information and communication technology (ICT) in healthcare, which has revolutionised patient care, data management, and service delivery. While ICT was initially used solely for administrative purposes, it is now broadly defined to include a range of information and communication technologies such as electronic health records (EHR), telemedicine and analytics that improve operational Efficiency, patient access and quality of care. The ability to innovate, such as AI, cloud computing, etc., provides real-time data access that helps healthcare professionals make better decisions and also improves patient outcomes. In particular, the paper showcases the government's initiative to create an integrated digital health system. The study highlights the need for strategic implementation of ICT to optimize health outcomes and availability and access to services, particularly in resource-poor settings.
Authors - Sachin Naik, Rajeshree Khande, Sheetal Rajapurkar, Kartik Dalvi, Shubham Rajpure, Vaibhav Kalhapure Abstract - SmartMail Insights is an intelligent web-based toolkit that is created for email management and all goes above delivering the basic functions of most online mail applications. Through the automation priority ranking, auto-responses and emails summarization, it makes it easier for the users to deal with urgency and important mails to emails that may not be very tiresome. The ML algorithms that it uses help easily sort the emails by content, sender, and, there are separate filters to highlight important emails with variable options. Auto replies are supported by NLP and there is the summarization of text to make it easier to read. Despite this, there are ways that SmartMail Insights could advance its current model of categorization one way is to incorporate its model for identifying and sorting through spam emails and promotional ones at least, into more refined sort of emails such as personal, business and so on since doing so would prove helpful in improving categorization accuracy
Authors - Shailesh Gahane, Payal Khode, Arya Kapse, Deepak Sharma, Pankajkumar Anawade Abstract - The accessibility for every type of user, including disability, ensures that the websites and applications are developed to allow the access of every user in this electronic world. For my research paper, I aimed to report important techniques and best practices for developing accessible websites and applications while researching the effectiveness of the established accessibility guidelines, the role of assistive technologies, and inclusive design strategies. The first objective of this research is concerned with the practical application and the effectiveness of the general core standards on accessibility overall, including the Web Content Accessibility Guidelines (WCAG) and the Americans with Disabilities Act (ADA), in terms of their positioning on promoting compliance and inclusion. Three are the targets of this paper. The first target concerns how assistive technologies, like screen readers and voice-control programs, interact with web applications along best practice recommendations for optimizing these tools to access better by following accessibility. Third is about inclusive strategies for design issues with color contrast, font selection, and responsiveness meant to improve accessibility for both visual, auditory, and cognitive impairment. This research gives a comprehensive and definitive understanding of the present techniques and best practices in accessible web and app development. Therefore, how the developers can possibly enhance usability and ensure digital inclusivity for all users is provided.
Authors - Madhuri Thorat, Priyanshu Kapadnis, Neel Kothimbire, Rameshkumar Choudhary, Atharva Jadhav Abstract - The emergency of Generative AI has led to the development of various tools that present new opportunities for businesses and professionals engaged in content creation. The education sector is undergoing a significant transformation in the methods of content development and delivery. AI models and tools facilitate the creation of customized learning materials and effective visuals that enhance and simplify the educational experience. The advent of Large Language Models (LLMs) such as GPT and Text-to-Image models like Stable Diffusion has fundamentally changed and expedited the content generation process. The capability to generate high-quality visuals from textual descriptions has exceeded expectations from just a few years ago. Nevertheless, current research predominantly concentrates on text generation from text, with a notable lack of studies exploring the use of multimodal generation capabilities to tackle critical challenges in instruction supported by multimodal data. In this paper, we propose a framework for generating situational video content based on English poetry, which is executed through several phases: context analysis, prompt generation, image generation, and video synthesis. This comprehensive process necessitates various types of AI models, including text-to-text, text-to-video, text-to-audio, and image-to-image. This project illustrates the potential of combining multiple generative AI models to produce rich multimedia experiences derived from textual content.
Authors - Rashmy Moray, Sridevi Chenammasetti, Shikha Jain, Ankita, Shivani Abstract - This study explores the determinants influencing the adoption of robo-advisory services among Generation Z and Millennials. Leveraging the DeLone and McLean Information Systems Success (DM ISS) model, the research examines four key dimensions—system quality, information quality, service quality, and user satisfaction—to evaluate their impact on users' intention to adopt these services. A structured questionnaire was utilized to collect primary data, which was analyzed using Structural Equation Modeling (SEM) via SmartPLS software. Findings highlight that service quality and user satisfaction significantly influence the adoption intent of robo-advisory services. This research expands the DM ISS model's application to robo-advisory services, providing valuable insights for stakeholders on how these dimensions contribute to user satisfaction and overall system performance.
