Skip to content
Case Study

Making sense of seabed imagery

Making sense of seabed imagery


The Agri-Food and BioSciences Institute (AFBI) is a leading provider of scientific research and services to government, non-governmental organisations and commercial organisations.

In 2022, AFBI sought a technology partner to deliver an Artificial Intelligence (AI) system for object detection with tracking capabilities for imagery of the seabed. The aim of this project was to provide an image annotation software to build a training set consisting of a variety of seabed objects and a platform to train and deploy an AI system for automaticobject detection, tracking, and counting during live capture of images. Historically, this process was completed manually and was therefore time and resource intensive.

The solution

Analytics Engines created a solution that utilises a range of advanced data-driven techniques and technologies including Artificial Intelligence and Computer Vision.

This solution allows users to upload video and images of the seabed and annotate an unlimited number of classes via a graphical user interface. This annotation platform includes semi-automatic annotation tools help to speed up the annotation process.

To train the Artificial Intelligence system, historic seabed image frames were annotated and used to train a deep-learning model for object detection. The model has been developed to support multiple object classes (e.g. nephrops, scallops, etc.) and can be updated or trained on a new class by simply providing additional annotated images showing the object classes of interest.

The solution includes a tracker function which can track objects of interest across video frames that provides users with a running count of objects detected of each type as well as confidence scores for the classified objects.

Share this article
by PJ Kirk

Fancy a chat?

Get in touch