A computer vision solution for nectar sugar estimation

Supervisor: Dr Petra Bosilj, Dr James Brown

What you will be doing 

Nectar rich habitats are a key driver of pollinator abundance and diversity, however the current in-field methods of analysis are labour intensive and are limited by the availability of a highly specialised workforce. 

We want to develop a computer vision solution for estimating the daily nectar sugar mass produced in stands of flowering vegetation. As the initial step, we will rely on a labelled public dataset to explore the capabilities and limitations of currently available solutions (Hicks, 2021) by exploring different options for the key components of model training (choice of backbone and architecture, hyperparameter tuning, augmentation techniques). 

Hicks, D., Baude, M., Kratz, C., Ouvrard, P. and Stone, G., 2021. Deep learning object detection to estimate the nectar sugar mass of flowering vegetation. Ecological Solutions and Evidence, 2(3), p.e12099.

What skills would be useful to have for this project 

We are looking for a student familiar with Python, with an interest in designing deep learning solutions to computer vision problems, and their applications in ecology. Any experience with training CNN models is a plus. The list of useful skills and interests includes: 

• Python 
• Machine learning frameworks (e.g. pytorch) 
• An interest in computer vision 
• An interest in applications in ecology and sustainability. 

How to get more information and apply 

For more information about the project, please contact Petra (pbosilj@lincoln.ac.uk) or James (jamesbrown@lincoln.ac.uk) by e-mail in the first instance. Please also send in your application, containing a brief description of your interests and your suitability for this project as well as a CV, jointly to both of the above e-mails.  

Candidates will be selected based on the content of the application, followed by a brief chat on Teams/in-person to confirm your suitability. The deadline for applications is 24/06/2024, with a planned project start 01/07/2024.