Authors - Sunil Kumar, Sanya Shree, Saswati Gogoi, Anshika Shreshth Abstract - The evolution of wireless communication, which led to the introduction of cellular networks, has enabled the highly interconnected world we experience today. This research paper discusses the developmental course from the first generation (1G), which introduced analog voice communication to the higher generations of networks, which brought digital signals into play. Multiple access technologies and significant emerging technologies of cellular networks from 1G to 5G are discussed. A comparative analysis among different generations of networks is presented, and a vision of the forthcoming sixth generation is presented. The role of the present widely used fifth-generation (5G) in defence, healthcare and education is discussed along with other applications. The challenges and future directions of the sixth generation (6G) Wireless Communication Network (WCN), which aims for ultra-low latency and extremely high energy efficiency using the specifications of artificial intelligence are discussed.
Authors - Shradha Naik, Suja Palaniswamy, Nicola Conci, Vishal Metri Abstract - Anomaly detection in videos from CCTV cameras can be an important strategy for crime analysis and prevention. The main focus of our work is on detecting the crime of chain snatching from videos captured in India. Due to the absence of a training set of similar Indian videos, it is challenging to design a classifier for this crime. Hence a technique called Model Agnostic Meta-Learning (MAML) is used to train a network on the well-known UCF crime dataset for detection of chain-snatching in a dataset custom built by us. MAML is further developed to result in a method called Sampling-based Meta-Learning Anomaly Detection (SMLAD). With this, the characteristics of MAML are used automatically to classify chain-snatching as an anomaly and obtain best accuracy and AUC scores of 86 % and 84 % respectively. Thus the proposed work demonstrates the efficacy of MAML to correctly classify chain-snatching which constitutes completely unseen data, as a crime-related anomaly.
Authors - Sanal Kumar S P, Arun K Abstract - Aspiring researchers have to consider choosing an appropriate Ph.D. subject. However, the complexity of the regulations and the large number of possible choices especially in context of cross and multi disciplinary approach render it challenging. The manual processing of applications by universities is time-consuming and prone to errors, which leads to inefficiencies and in-ordinate delays. We created DSPredict, a novel approach that employs machine learning to identify the most appropriate Ph.D. subject for each applicant. Our methodology assesses application profiles and predicts the most suitable subjects. The findings suggest that DSPredict surpasses traditional methods, resulting in increased accuracy and significantly shorter time to identify appropriate subjects.
Authors - Sudha S K, Aji S Abstract - Rapid advancements in video surveillance and analysis require advanced frameworks capable of detecting, segmenting, and tracking objects in complex, dynamic scenes. This paper introduces DySAMRefine, a novel dynamic scene adaptive mask refinement strategy for robust video object segmentation and tracking (VOST) in dynamic environments. DySAMRefine is built upon a Mask R-CNN pipeline for instance-level segmentation and incorporates a long short-term memory (LSTM) network to capture temporal dependencies, ensuring smooth and consistent object tracking across frames. A spatio-temporal attention block (STAB) is introduced to maintain temporal coherence, supported by a temporal consistency loss (TCL) that penalizes abrupt changes in masks between consecutive frames, promoting temporal smoothness. DySAMRefine dynamically adjusts mask refinement based on the complexity of the scene and optimizes performance in static and highly dynamic environments through a deformable convolutional network (DCN). The training process employs an efficient mixed precision scheme to minimize computational overhead, enabling real-time performance without sacrificing tracking precision. Extensive experiments and ablation analysis demonstrate that DySAMRefine enhances the accuracy and robustness of VOST, achieving superior J&F scores on benchmark datasets.
Authors - Roshan Kamthe, Yash Gaikwad, Shubham Pawar, Kishan Chandel, Pushpavati Kanaje Abstract - The purpose of this research is to develop a system that can identify hand movements, facilitating easier communication for the deaf and mute. Apart from providing voice output for calls coming in from non-deaf individuals, the system also includes a mobile application that allows users to communicate through hand gestures. Our solution gives those who are hard of hearing or deaf a straightforward way to communicate by utilizing modern technologies like computer vision and machine learning. The goal of this project is to develop a hand gesture detection system that will improve communication accessibility for people with speech and hearing problems, especially the deaf and mute community. Our project's primary objective is to provide individuals who are incapable a clear solution.
Authors - Atharva Desai, Anurag Raut, Aditya Thatte, Ramchandra Mangrulkar Abstract - This project proposes an advanced, multi-threaded, opensource NoSQL database architecture designed to extend and improve upon existing database systems. The architecture utilizes a shared-nothing approach, sharding the keyspace into multiple parts, each managed by a dedicated thread. By employing hash-based ownership, the need for synchronization is eliminated, thereby reducing performance bottlenecks. The system is optimized for distribution within a single machine, leveraging a thread pool technique to manage potential thread overhead efficiently. Additionally, the database replaces traditional hash tables with dashtables, which minimize rehashing overhead and optimize memory usage by segmenting the hash space into smaller, more manageable portions. This novel approach significantly improves efficiency and scalability, providing a compelling alternative to existing solutions like Redis.
Authors - Trupal J. Patel, Mahek D. Viradiya, Jaykumar B. Patel, Dhruvi J. Patel, Prisha M. Patel, Dhruv Dalwadi Abstract - In the era of surging urbanization, the problem of managing waste effectively has become a major concern. This research paper provides a solution by providing a cutting-edge system for real-time monitoring and management of waste bins using IoT sensors integrated with cloud computing technologies. By using an ultrasonic sensor (HC-SRO4) to precisely and accurately gauge levels of waste with a DHT22 sensor to monitor conditions related to the environment. This solution provides innovation that enables the data collected precisely to enhance the efficiency of waste management. The data that is collected is then processed by a Raspberry Pi, which is the core unit of the whole system, that transmits the whole information to a cloud platform where analysis and visualization are done. This makes it possible for stakeholders to access real-time insights of waste levels and factors affecting the environment, which constantly improves the process of decision-making. Moreover, the system integrates predictive analysis to predict waste collection trends, enabling the optimization of collection schedules and minimizing the trips that are unnecessary for collection. By this way, the operational cost can be reduced, and it helps in improving the efficiency of service. This approach not only considers logical challenges but also serves sustainable waste management practices. Ultimately, this research illustrates the potential of IoT technologies to transform, creating smarter and more adaptive environments in urban areas.
Authors - Divyashree HB, Deepthi Chamkur V, Preesha Tandon, Laranya Subudhi Abstract - In today's technology age, a Ground Control Station System application software for offboard mode and manual control of unmanned aerial vehicles is essential for a variety of onboard activities like tracking, surveillance, and patrolling. This study discusses software that controls and collects important data from unmanned aerial vehicles. The program is developed in Python 3, and the graphical user interface is created with the Qt5 framework. Melodic is the robot operating system (ROS) that facilitates communication and networking. The software allows you to control the drone's forward, backward, up, down, left, and right motions. The live feed from the RGB camera (day camera) and the night vision camera may be watched and saved as snapshots. It is also possible to save the live stream footage to a CD. Object tracking and detection functions are offered for surveillance purposes. The software may also be used to operate a gimbal fitted to the drone. The entire program is beta tested on the Gazebo real-world simulation, and the experimental findings are based on a real-world hexacopter flight.
Authors - Sandeep M.Chaware, Mohit Matte, Pratik Dahagaonkar, Anurag Deotale, Laukik Pagar, Jayesh Sarwade Abstract - Agriculture is the land-cultivation, crop-growing, livestock-raising processes. A nation's economic growth depends on its agricultural sector. Agriculture makes for about 58% of a nation's primary revenue source. Up to now, farmers sow and cultivate or practice agriculture based on favorable weather and soil conditions without considering the future supply and demand of crops and the type of agriculture practiced, thus often doing reduce profits from agriculture. Typically, when demand for a crop is low and supply is high, the price drops too low, leading to debt for the farmer and vice versa. Predicting what crops should be grown or what type of agriculture should be adopted in today's world is essential to meet people's needs and increase farmer productivity. Machine learning, data mining, and data analytics can be used to collect data, train models, and predict the market demand, supply chain, demanding type of agriculture and location of agriculture for revenue generating agriculture. This will help reduce losses for farmers. Due to the ongoing changes in the world, the proposed Machine Learning assistanat helps determine how to manage agriculture intelligently. It assists an individual towards profitable agriculture This work's primary goal is to sustain a single farm profitably while achieving high output at reasonable expenses. Questions including pricing comparisons, government activities, plant protection, animal husbandry, weather, and fertilizer management are addressed by the proposed method.
Authors - Priyanshi Desai, Parth Shah Abstract - The increasing adoption of voice-controlled IoT devices, such as Amazon Alexa, Google Home, and Apple’s Siri, has transformed modern interactions with smart systems in various sectors, including home automation, healthcare, and industry. While these devices offer convenience and enhanced accessibility, they are also vulnerable to significant cybersecurity threats. This paper examines the security challenges associated with voice-controlled IoT systems, focusing on key vulnerabilities such as voice spoofing, man-in-the-middle attacks, insecure APIs, and data privacy concerns. Additionally, the paper explores various attack vectors, including adversarial attacks and physical tampering, and assesses current mitigation techniques like biometric voice authentication, secure data transmission, and anomaly detection. Privacy concerns are also discussed, particularly in relation to data retention and third-party access. As the use of these systems continues to grow, advanced cybersecurity measures, including quantum-resistant encryption and enhanced biometric methods, are essential for securing voice-controlled IoT devices. Furthermore, the establishment of regulatory frameworks to govern the handling of voice data is critical. This paper concludes by identifying future directions to improve the security and privacy of voice-controlled IoT devices, emphasizing the need for innovative solutions to counter an expanding array of cyber threats.
