High throughput Pacific oyster phenotyping using computer vision and machine learning

Supervisory Team:

Primary supervisor: Dr Dean Giosio

Co-supervisor: Dr Andrew Trotter

Additional supervisors: Prof. Byeong Kang (ICT – to be confirmed)

Brief project description:

This project is a joint investigation with Australian Seafood Industries (ASI) to improve breeding lines for uniformity of shell shape and growth of the the Pacific oyster, Crassostrea gigas.

This project aims to develop a suitable analysis pipeline for performing large-scale physical assessments of oyster shell charateristics using computer vision and machine learning techniques.

The developed model will be used to answer the question: ‘do certain breeding lines exhibit more uniform shell development compared to other lines, when grown in comparable environmental conditions?’ If so, how can this information be quanified and utilised for family selection based on traits relating to shell uniformity.

Ultimately, the developed model will provide growers with a tool for performing high-throughput visual phenotyping of the Pacific oyster. Access to such data would enable family lines to be selected based on desired physical traits, improving farm efficiency, and providing a more consistent product to market.

Skills students will develop during this research project:

The student will be responsible for managing the project and will therefore develop important skills in project planning, industry consultation, time management, and scientific report writing. Technical skills will be developed in the areas of experimental program design, data collection, various machine learning methods (image processing, object segmentation using neural networks, k-means clustering), and in large dataset handling and analysis.

This project will develop skills that are highly transferrable and applicable across many industries.

Authorised by the Executive Director, Institute for Marine and Antarctic Studies
January 27, 2022