/ROVer-Optometry-Underwater-Object-Detection

The members in this team will learn the basics of image classification to then be immediately thrown into the ringer to train a model that can detect certain types of objects under various measures of turbid water by utilizing CNNs (Convolutional neural networks) such as YOLO or MobileNet.

ROVer Optometry: Underwater Object Detection

Introduction

Understanding ROVs

ROV, short for Remotely Operated Vehicle, is a crucial tool for various underwater activities. Designed to operate in challenging aquatic conditions, ROVs are remotely controlled, allowing for tasks and data collection in areas that are often inaccessible or dangerous for humans. RoboSub @ UTD's VESPA ROV

RoboSub @ UTD's VEPSA ROV

Problem Focus

In this project, team members will embark on a journey to master the fundamentals of image classification. Following this foundational knowledge, the team will dive into the realm of training models using advanced convolutional neural networks (CNNs), including YOLOv7. Their mission: develop and fine-tune a model capable of detecting trash beneath varying levels of turbid water. The ultimate objective is to showcase the achievement of precise and high-accuracy underwater object detection and classification.

Dataset

To create our main dataset, we gathered data for a multitude of places. Which includes, RoboFlow, a site where others can upload their own datasets to conform to many things

Model

Training

Results

Analysis

Conclusion

Contributors