Creating a model to recognise and classify boulders on the seabed.
The team at the Hartree Centre North East Hub worked with Ternan Energy to develop a detection model to aid with the identification and classification of boulders on the sea floor to aid the development of offshore windfarms.
Ternan Energy is an innovative independent marine geoscience consultancy providing geoscience professional insights. One of the solutions Ternan Energy provides involves lending their expertise to offshore windfarm development, whereby they are required to manually identify and classify regions of the sea floor from side-scan sonar data. The goal of the project was to automated models that could reduce the time and cost involved in their current manual work process.
We worked with Ternan Energy to explore the production of a proof-of-concept framework to develop bespoke deep learning models that can be applied to side-scan sonar mosaic data (images of the sea floor). The initial stages involved exploring current literature in the domain and where others have explored neural networks applied to similar geoscience problems. We created code to ensure groups of sliding window images were split properly whilst undergoing training and validation. We created a modular extensible framework using PyTorch to enable experimentation and tuning of hyper-parameters involved, tracking these experiments using MLFlow, with streamlined data preparation pipelines to handle complex hydroacoustic data. We produced detection models from these experiments that could be applied to aid the detection of boulders of varying sizes on the seafloor.
The code deliverable and report produced by Hartree Centre North-East Hub will enable Ternan Energy to begin further iterating and improving on the detection models produced with their own client data. As a result of the project, Ternan Energy are looking to begin aiding their manual processes and improving upon the models with further data.
"The Hartree Centre North East Hub has provided us with a valuable framework and AI model which we can improve with further data. Through the sessions, we have been able to explore and scope a tool which will speed up our seabed boulder detection processes. This will allow us to provide faster and more reliable geoscience solutions for our clients."
- Ternan Energy
This work was completed as part of one of our collaborative data projects. The projects are up to 12 weeks in duration and give you access to a wide range of expertise across our team of data scientists and data engineers. We will work alongside your team to scope your data science or engineering project, build a prototype solution, and explore options to deploy it within your organisation. You can learn more about them on our webpage here.
If you would like to learn more about the Hartree Centre North East Hub or our collaborative data projects, please get in touch with us at: hello@hartreenortheast.uk