IT - Toc H Institute of Science & Technology

Student Projects

Academic Projects in Information Technology

2025-2026

POTHOLE DETECTION AND MONITORING IN AUTONOMOUS VEHICLES USING MACHINE LEARNING AND IOT
TEAM MEMBERS: Amnah Yasir, Gayathri Vinodh, Nasih Rahman K A GUIDED BY: Asst.Prof. Nifasath Piyar S
ABOUT PROJECT: Based on YOLOv8, a sophisticated convolutional neural network created especially for object detection, this work offers a real-time, non-invasive system for autonomous navigation and pothole detection. By precisely identifying potholes, which pose a serious risk to drivers, the system seeks to enhance road safety and promote the development of affordable intelligent transportation systems. In order to integrate computer vision, machine learning, and sensor modules for efficient real-time navigation and detection, the system architecture uses a Raspberry Pi as the main computing component and an Arduino as a secondary microcontroller. A Raspberry Pi camera, which performs essential tasks like object detection, traffic signal identification, and lane detection, is a key component. For avoiding obstacles and determining the depth of potholes, ultrasonic sensors are crucial. The YOLOv8 model can simultaneously generate bounding boxes as well as recognize identified objects because it was trained on a sizable sample of 720×720 pixel photos that represented a range of realworld environmental settings. Both fully autonomous and human-driven cars can use the model, which supports road maintenance procedures and gives drivers visual alerts in real time.
SDGs MAPPED: 3,8,9,11,13

A SCALABLE AND SECURE CRYPTOGRAPHIC SOLUTION FOR RESOURCE-CONSTRAINED IoT DEVICES
TEAM MEMBERS: Adarsh Jiju David, Alan James Robert, Aswin Mammen, Eldhose Babichan GUIDED BY: Asst.Prof. Sandhya C P
ABOUT PROJECT: This project focuses on the design and implementation of a lightweight cipher algorithm for secure communication in resource-constrained environments such as IoT and RFID systems, addressing the growing need for security in modern connected devices. Many existing lightweight encryption techniques either consume high computational resources, lack resistance to advanced attacks, or are inefficient for software implementation, while some also suffer from limited data handling due to small block sizes. To overcome these challenges, a novel lightweight encryption algorithm is proposed with a focus on security, performance, and cost-efficiency, making it suitable for devices with limited processing power and memory. The algorithm was implemented on Xilinx Vivado and evaluated through various testing and validation methods to ensure reliable encryption and decryption, efficient performance, and resistance to security threats. The results demonstrate that the proposed approach provides effective data protection with fast processing suitable for embedded systems. This system can be applied in RFID-based automated toll plaza systems to ensure secure communication between RFID tags and readers, preventing unauthorized access and cloning attacks, and can also be extended to applications such as smart transportation, healthcare systems, defence infrastructures.
SDGs MAPPED: 3,8,9,11

Skull2Face: Anatomy Guided 3D Facial Recoustruction System Using Deep Learning and Tissue Depth Modeling
TEAM MEMBERS: Aiswarya Michael, Akash Reji, Anjali Raju, Serin Mariyam Varghese GUIDED BY: Prof. (Dr.) Rasmi P S
ABOUT PROJECT: 3D Facial Reconstruction from the skull is very important in forensic, anthropological research. Traditional reconstruction approaches are manual, subjective, and time-consuming. Skull2Face introduces a complete end-to-end data-driven anatomy-guided framework to reconstruct 3D faces from skull images, encompassing biological profiles and statistical tissue depth models while providing anatomically plausible results. A combination of state-of-the-art diffusion-based image synthesis, 3D deep learning modeling and facial landmark optimization allows our system to synthesize realistic human faces from skull images inputs. The system produces identity-consistent, anatomically accurate, and visually-realistic reconstructions using anatomical landmarks, biological profiles, and generative diffusion models for more adaptable personal matches. The method we propose applies anatomical landmarks with statistical modeling for accuracy while still providing a detailed, individualized, reconstructive outcome with realistic texture. It produces realistic facial textures with a high level of visual authenticity by using generative diffusion models. Described as a modular method, this framework helps more limited dependence on large datasets, representing a more rapid, individualized, and automated alternative methodology to facial reconstruction. Compared to traditional or artist-driven reconstruction techniques, this system provides a rapid, easy-to-use, and repeatable alternative. This refined approach enhances realism, and accessibility in forensic reconstructions while minimizing the computational costs.
SDGs MAPPED: 3,4,9

2024-2025

Waste segregation using Robotic arm using AI
TEAM MEMBERS: Abhisheik Muralidharan Nair, Amina Ahmed, Rony Mathew Antony, Roshan Mathew Varghese GUIDED BY: Asst.Prof. Kurien Thampy
ABOUT PROJECT: This project presents an automated system for waste segregation utilizing a combination of AI, computer vision, and robotics. The proposed system employs the YOLOv8 object detection model, which efficiently classifies multiple types of dry waste, including paper, plastic, metal, and glass, in real time. Integrating this model with a 5-degree-of-freedom robotic arm, the system performs targeted waste segregation by directing different waste types into designated bins. This approach not only enhances segregation accuracy but also automates a process traditionally reliant on human labour, thus reducing health risks and improving operational efficiency in waste processing environments. The system’s architecture features a Raspberry Pi that processes live video streams from an external camera, identifying waste items within the frame. The YOLOv8 model, trained on a comprehensive dataset, accurately classifies waste categories and transmits commands to an Arduino microcontroller that controls the robotic arm. The robotic arm, equipped with precise servo motors, then physically sorts the waste items according to their classifications. During experimental testing, the system demonstrated an average classification accuracy of 80% in a controlled environment, with plastic and glass achieving the highest precision. The incorporation of both hardware and software components ensures an optimized balance between computational power and physical manoeuvrability, making this system a practical and scalable solution for modern waste management challenges. Future work aims to refine the model’s ability to classify complex waste types, particularly paper, by expanding the training dataset and enhancing processing power. This project underscores the significant potential of AI and robotics to revolutionize waste management systems, contributing to a cleaner environment and more efficient resource recovery. The study advocates continued research and development in AI-driven waste management systems to support environmental conservation and promote sustainable waste practices
SDGs MAPPED: 3,4,8,9,11

Weed Detection and Removal using Laser Beams
TEAM MEMBERS: Bobby Subhash, Draupadi S Krishna, Sana Alex GUIDED BY: Asst.Prof. Neomi Nelin Nicholas
ABOUT PROJECT: Introducing an innovative laser-based weed detection and removal system designed to revolutionize sustainable agriculture. Using high-quality imaging, the system accurately distinguishes between weeds and crops in real-time, allowing for precise targeting and removal of only unwanted weeds. High-resolution cameras and sensors capture detailed, real-time images of the field, which are processed by advanced machine learning algorithms to identify weeds with high accuracy. This enables the system to differentiate crops from weeds, ensuring that only the weeds are targeted. Once identified, calibrated laser beams are deployed to target and disrupt the cellular structure of the weeds, effectively destroying them without causing harm to nearby crops. This laser-based approach eliminates the need for traditional weed management practices that often degrade soil quality, such as heavy herbicide use or soil-disturbing mechanical weeding. By focusing on weed elimination at a cellular level, the system leaves the soil structure intact, preserving soil health and supporting long-term sustainability in agriculture. In addition to promoting environmental benefits, this system significantly lowers weed management costs by reducing the need for manual labor and minimizing dependency on herbicides. Farmers can thus manage weed growth more efficiently, reducing operational costs and enhancing crop yield. Overall, this advanced weed detection and removal system provides an effective, eco-friendly solution that supports sustainable farming practices, conserves soil integrity, and offers a more efficient and scalable approach to managing weeds in large agricultural fields.
SDGs MAPPED: 2,3,4,8,9

Intelligent IOT based River Pollution Monitoring
TEAM MEMBERS: Ashly Jayaraj, Nandana Mol R, Naresh Ramesh, Titus Alex GUIDED BY: Dr. Rasmi P S
ABOUT PROJECT: An IoT-based smart water quality monitoring system enables the efficient real-time tracking of important water parameters like pH, turbidity, and temperature. This is significantly different from the old traditional systems, which were slow, manual, and limited to providing only very limited data; it applies low-power sensors that collect and continuously transmit data through networks like Wi-Fi to a cloud-based platform. The data is safely stored and analyzed, which further provides actionable insights through advanced analytics and a machine learning algorithm. Water quality can be monitored using intuitive dashboards and water quality parameters can send out real-time alerts when the values go beyond safe levels. The system is highly cost-effective, which means it reduces the expensive need for manual sampling and laboratory testing, while offering scalability for large or remote areas. Its low power consumption allows for long term operation without frequent maintenance. It is thus suited to mass deployment in various settings, such as environmental monitoring, water treatment plants, industrial waste management, and agriculture. The system helps protect public health through timely interventions and prevents waterborne diseases, besides ensuring that industries are in compliance with regulatory measures. Its real-time capability allows the authorities and organizations to respond rapidly to water quality issues, thereby improving water management and resource sustainability.
SDGs MAPPED: 3,4,6,9,11

2023-2024

Autonomous Water Hyacinth Detection and Removal System for Water Bodies
TEAM MEMBERS: Parvathi S, Mahima Susan Rajesh, Athira M V, Emmanuel Rajan GUIDED BY: Asst.Prof.Sandhya C P
ABOUT PROJECT: Water hyacinth, an invasive aquatic plant species, poses a significant threat to the ecological balance and economic activities in aquatic ecosystems worldwide. This idea aims to design, develop, and deploy an autonomous Unmanned Surface Vehicle (USV) equipped with cutting- edge technology, including computer vision and artificial intelligence, to efficiently combat water hyacinth overgrowth. The USV is designed to identify, collect, crush, and heat water hyacinth, making it an innovative solution for environmental restoration and preservation. The system study encompasses a thorough assessment of technical requirements, stakeholder expectations, environmental impacts, and regulatory compliance. The USV integrates advanced computer vision algorithms to identify water hyacinth patches accurately, ensuring targeted removal while minimizing harm to other aquatic life. An autonomous navigation system facilitates obstacle avoidance and precise control during mission operations. Additionally, the system considers environmental sustainability through rigorous impact assessments and safety protocols. The USV’s collected water hyacinth undergoes onboard processing, including crushing, compressing and heating, facilitating preprocessing and disposal. Through collaboration with local communities, environmental agencies, and potential users, this project seeks to raise public awareness, foster collaboration, and address the multifaceted challenges posed by water hyacinth infestations. The system is integrated with solar-panel to reduce the use of fossil fuel and to make it cost effective. The successful implementation of this project holds the potential to mitigate the adverse effects of water hyacinth overgrowth, protect aquatic ecosystems, support fishing and farming activities, and serve as a model for sustainable environmental management.
SDGs MAPPED: 3,4,6,9,14

Hybrid Blockchain -AI Framework for Adaptive Flood Monitoring and Early Warning System
TEAM MEMBERS: Preema S, Anamika C J, Andria P.V, Adiq P N GUIDED BY: Asst.Prof.Smitha Mohan
ABOUT PROJECT: The “Hybrid Blockchain-AI Framework for Adaptive Flood Monitoring and Early Warning System” project is aimed at revolutionizing flood prediction and management in response to the escalating frequency of floods worldwide. At its core, the framework integrates a diverse range of sensors, including those for water level, humidity, temperature, and pressure, to capture a comprehensive set of environmental parameters crucial for flood monitoring. Through the utilization of Python programming language, sophisticated machine learning algorithms such as Support Vector Classifier (SVC), Random Forest, and Decision Tree are implemented to analyze both historical trends and real-time sensor data. This analytical approach enables the system to generate precise flood predictions, thereby facilitating timely responses to potential flood risks. A key aspect of the framework is its ability to deliver alert messages to registered users through mail interface, empowering communities with the information needed to take proactive measures in preparation for impending flood events. By seamlessly integrating advanced technologies and data-driven insights, this project seeks to enhance flood resilience and mitigate the adverse impacts of flooding on vulnerable populations and infrastructure
SDGs MAPPED: 3,4,6,9,11

Dark Web Cryptocurrency Transaction Monitoring for Illicit Drug Trade Detection with Bio-Inspited Object Recognition
TEAM MEMBERS: Bestin Thomas Biju, Blesson Reji Varghese, Praseed Sreepadmakumar, Athul M P GUIDED BY: Dr. Rasmi P S
ABOUT PROJECT: The proposed system is designed to combat illicit drug trafficking on the dark web by leveraging transaction IDs from the blockchain and incorporating Bitcoin currency conversion. Through the application of data analytics and deep learning techniques, the system aims to analyze Bitcoin transactions associated with known darknet markets, effectively identifying patterns indicative of drug trafficking. This innovative approach facilitates the cataloging of active and legitimate darknet markets involved in drug sales, along with the identification of individuals engaged in drug trafficking within India. The system’s comprehensive methodology involves the collection and analysis of transaction data, including associated account and server hosting details, providing law enforcement agencies with valuable insights to apprehend those involved in drug-related activities. The development of this system spans multiple stages, including data collection, processing, and the training and testing of deep learning algorithms using available datasets. Model building and implementation, coupled with performance analysis, form integral components of this initiative. It is imperative to underscore that, throughout this process, strict adherence to applicable laws, regulations, and privacy considerations is paramount. Ensuring compliance with legal and ethical standards is integral to the responsible development and implementation of the system, contributing to a proactive and effective strategy against drug abuse in the country.
SDGs MAPPED: 3,4,9,16

2022-2023

Cryptocurrency price prediction using sentimental analysis and Machine Learning
TEAM MEMBERS: Alen Saju, Aravind Madhu, Adnan Irshad, Arjun V GUIDED BY: Asst. Prof. Neomi Nelin Nicholas
ABOUT PROJECT: The popularity and importance of cryptocurrency, along with its relevance, has been growing in the recent years. Many organizations throughout the world have acknowledged the importance of adapting to this technology to avail the various numerical benefits virtually, with rapid advances seen in this direction. Price prediction happens to be one of the biggest challenges in today’s financial market situation, as the dilemma to buy or sell cryptocurrency is an interesting challenge faced by traders. Since the past few years, it has reached unparalleled highs leading to theories explaining the trend in its growth. The idea of whether the cryptocurrencies’ movements and momentary fluctuations can be predicted, has kept investors, researchers and other economists very occupied as of late. In order to make these decisions in the crypto market, it is necessary to have efficient and reliable tools of price prediction, profitability and risk assessment, at least for the short- term time range. We can achieve this by using machine learning models such as LSTM (Long Short-Term Memory), which can learn long term dependencies. The key factors used are available price, close price, high price, low price, volume and market cap with the interdependencies amid some cryptocurrencies, thus centers on evaluating vital features that influence the trade’s unpredictability by applying the model to increase the effectiveness of this process.
SDGs MAPPED: 3,4, 8,9

Blockchain Technology to Link Road, Gas & Electric Works
TEAM MEMBERS: Siddharth Santhosh, Georgy Boban, Jefin Jose, Ebin Joseph K GUIDED BY: Asst.Prof.Smitha Mohan
ABOUT PROJECT: The development of Blockchain technology in e-government still needs discussion in different aspects; this technology offers a new method for delivering and managing public services, and there remains a need to establish standards, deploy solid management systems and ensure adequate security to make sure the services and platform are reliable, authoritative and supportive of long term preservation. Blockchain has the potential to change Indian society in many aspects. Its development still has both opportunities and risks. However There is still a need for Blockchain companies and market administrators to actively collaborate with each other, implement Blockchain operations, and introduce innovative solutions. Therefore our project introduces a new idea which supports the government in the field of services like water ,Gas, Electric work by collaborating them using Blockchain technology.
SDGs MAPPED: 3,4,9,11

Deep Learning based software solution using bio- inspired objects to detect illicit drug trafficking in dark web
TEAM MEMBERS: Steev John, Devika J Nair, Priyanka.Prince GUIDED BY: Dr. Rasmi P S
ABOUT PROJECT: The Dark Web is the darkest and most anonymous part of the Internet, where many illegal activities, including illicit trafficking in drugs, take place to a very large extent. Tracking of such activities is quite difficult from a law enforcement perspective given the inherent anonymity and volume of content on the dark web. Currently there is no efficient system to detect and monitor such activities in dark web. So, the concept of utilizing data analytics and deep learning techniques for making an intelligent system which can crawl through the dark web to catalogue active and genuine darknet markets selling drugs, identify drug traffickers on dark net based in India and the drugs they offer for sale and collect the account and server hosting details of the drug traffickers across the Dark Web. Furthermore, these findings are handed over to the law enforcement. Hence, this leads to the curtailing of drug abusers in our country and turns our nation into a drug – free motherland. Development of the proposed System includes different steps like Data collection, Data processing, training and testing Deep learning algorithm using already available datasets, model building and implementation, designing an efficient dark web crawler, performance analysis etc.
SDGs MAPPED: 3,4,9,16