COMP530 - MSc Group Project

Using Deep Convolutional Neural Network to Detection Bio-energy Image Segments

The aim of this project is to develop a desktop application that is capable of detecting bio-energy sources within images. It is developed in Python with PyQt and TensorFlow.

Keep track of the project progress with the Kanban Board

Technical documentation can be found at docs/technical.md and user documentation can be found at docs/user.md.

Setup Guide

Ensure you have Git LFS and TensorFlow installed. The project is written in Python 3.7.

Clone the project: git clone https://github.com/Delphboy/COMP530.git

Navigate to the project root: cd COMP530

Install dependencies: pip install -r requirements.txt

Running The Project

The project can be operated from a UI or via the command line via a CLI.

• Recommended: Running the UI: python src/main.py --gui
• Running an existing model on the dataset via the CLI: python src/main.py --detect <location of the model .h5 file> <directory of training set>
• Training a model via the CLI python src/main.py --train <directory of dataset> <train ratio> Where the train ratio is: 0 < train ratio < 1.

Contribution

• Any PRs to the master branch require a code review from one other team member
• Any PRs will undergo automated linting checks via GitHub Actions. These checks must pass before the PR can be merged.

Team Members

• Bharathkumar Ayyadurai
• Sai Saranya Donthukurthi
• Guo Li
• Zhiying Li
• Susanthi Mandalapu
• Henry Senior