DCU Expo 2024 - Final Year Projects

101 213. Fruit Classification UsingDeep Learning Computer Vision Utilising deep learning to develop a classification algorithm to classify various fruits. The model employs established computer vision practices like feature extraction and newer technologies like residual connections (e.g., ResNet50V2), striving to create a robust model capable of achieving high accuracy. Through extensive research and implementation, optimised key aspects including data preprocessing, model architecture design, and hyperparameter tuning are applied to enhance the performance of the deep learning algorithm. The fruit classification algorithm holds significant potential for applications in agriculture, food processing, and dietary monitoring. Student Programme Mechatronic Engineering (Year 5) Project Area Artificial Intelligence, Computer Vision, Data Analytics, Image/Video Processing, Software Development Project Technology Python, Machine Learning, Deep Learning Student Name(s) Faruq Sulaimon Email faruq.sulaimon2@mail.dcu.ie Supervisor Prof Paul Whelan 214. Live Scene Annotation with TensorFlow The goal of this project was to create an object detection model suitable for real-time annotation of captured footage to be deployed in a low-impact embedded device to speed up annotation pipelines in the automotive industry. This model was created in Google Colab using the TensorFlow library to detect common classes in automotive machine learning applications, cars, pedestrians, road signs, traffic lights and bicycles. The model was deployed on an NVIDIA Jetson Orin Nano connected to a front-view external vehicle camera to simulate being deployed in a vehicle. This device can be deployed in a capture vehicle to create timestamped annotations of captured footage allowing the annotation teams to find desired classes within large, recorded footage samples much faster. Student Programme Electronic and Computer Engineering (Year 4) Project Area Artificial Intelligence, Automotive Technology, Computer Vision Project Technology Python, Machine Learning, Object Detection Student Name(s) Conor Delaney Email conor.delaney59@mail.dcu.ie Supervisor Prof Paul Whelan 215. Virtual Commissioning of a Three Pneumatic Cylinder System This project aims to twin the operation of a three pneumatic cylinder printing press using the simulation software Simumatik. This is achieved through real-time communication with the PLC of the physical printing press, and leveraging Simumatik’s physics capabilities. This Digital Twin Instance (DTI) is a true representation of the systemand can be used to facilitate two-way communication or can be run in a “headless” mode to enable experimentation or to implement changes to the process. Student Programme Mechatronic Engineering (Year 5) Project Area 3-DModelling, Automation, Cloud Computing, Educational, Internet of Things, Robotics, Simulation, Software Development, Human-computer Interaction Project Technology PLC Programming, Python, Solidworks, XML Student Name(s) Eoghan Craven-Grace Email eoghan.cravengrace5@mail.dcu.ie Supervisor Dr Nigel Kent

RkJQdWJsaXNoZXIy MTQzNDk=