Scottish Marine and Freshwater Science Vol 11 No 18

Underwater video is widely used to collect information on potential impacts of marine renewables on the environment being non-invasive, autonomous and cost-effective. However, the time intensive nature of assessing the video footage often means only a proportion of it is used.

This project provides an overview of the current state of computer vision technologies for automated detection of aquatic life in underwater video and provides a development route for a tool to analyse the large amount of video footage without the need for human supervision.

The project successfully demonstrated that image analysis techniques could be used on Marine Scotland Science video to a) distinguish salmon from sprats in a fish trap inside a fish trawl; and b) identify the phosphorescent sea pen from video that was generated while towing a video camera along the seabed.

The application of such automated image analysis technology will ultimately provide a fast and accurate method for identifying fish and benthic fauna from video. Specifically, it will enable a better understanding of movements of salmon around offshore wind farms and identify features of conservation interest in benthic video.