Coursework Research Project (2022)
NOTE: Please send this form to [email protected] by 24th January’ 2022.
Project Title Automatic object (bird) recognition from images with machine
learning
Supervisor(s) Aaron Spence (School of IT), Prof Marcel Klaassen (School of LES)
Email [email protected]
Campus Project can be conducted remotely, so we can accept any student
Start Date ☒ T1-2022 ☐ T2-2022 ☐ T3-2022
Project Description Project Description – [Not more than 300 words]
As part of our ongoing project titled Birdmark (www.birdmark.net),
observers can submit details of sighted shorebirds into the Birdmark
website, which are automatically vetted (via R scripts) and then included
into our database.
However, sometimes observers want to upload an image of sighted
shorebirds instead (see Appendix A). Uploaded images must be manually
processed which is quite cumbersome, and we hope to automate part of the
process.
Problem statement:
- Automated detection and marking of birds within an image
Methodology:
- Evaluate the accuracy of different machine learning frameworks
- Propose and demonstrate the most suitable option with test data
- Develop a tool that can automatically mark any detected birds (e.g.
place a rectangle around each bird)
Note that for this project we do not need to identify the species of the bird,
although we would welcome such functionality if possible.
Keywords: machine learning, deep learning, image processing, python
programming, Matlab
Student Necessary Skills - Moderate programming skills, with a willingness to learn Python (or
Matlab).
- A basic understanding of machine learning and/or image processing
techniques.
- A willingness to research the topic and literature
Below fields are for administrative purposes only (not to be shared in the handbook)
Hardware and Software Please mention if your project requires any specific hardware or software
Resources Required which need to be made available. Also mention if you already have access to
the required resources.
How many students can This project would best suit an individual student; it is not large enough to
this project support? warrant a team of students.
Rationale (if any)
Appendix A
