Personal tools

Skip to content. | Skip to navigation

Sections
You are here: Home / Our People / Students / Adam Gobi

Adam Gobi

Automated Benthic Species Mapping using Intelligent Computer Vision

Adam Gobi
Current image JPEG image — 30 KB

Combined the experience gained from a B.Eng. in Electrical Engineering and a M.Sc. in Artificial Intelligence into a Ph.D. in Ocean Engineering, concentrating on underwater robotics and machine vision. Recently assumed a role as lead camera engineer for James Cameron’s Deepsea Challenger submersible, which included a sixty-day expedition on the South Pacific Ocean to explore the deepest places on earth. Now back in Newfoundland setting up commercial R&D operations, continuing to fuel a passion for creating and deploying advanced technology for ocean exploration and protection.

Kim Juniper, Ray Gosine, Ralf Bachmayer

Faculty of Engineering and Applied Science, Memorial University of Newfoundland

9/2007 to 12/2012

Marine Biodiversity, MB-08: Image Analysis and Object Recognition Tools

afgobi@mun.ca

Many of the projects within CHONe include painstaking visual analysis of the seabed by sifting through large amounts of video and/or photo data.  This project aimed to develop a software tool to help automate part of that process, through the creation of advanced artificial intelligence and computer vision algorithms that are able to learn how to identify and count various benthic organisms appearing in seabed imagery.

The project was a success, and has been proven on two real-world data sets:

• Counting deepsea crabs in imagery collected by the Remotely Operated Platform for Ocean Sciences (ROPOS) ROV off the coast of BC, Canada 

• Counting sea urchins in imagery collected by the Sirius AUV off the coast of Tasmania, Australia.

Accuracy levels hovered around the 80% mark, in a way that was reliably repeatable.

Gobi has recently formed an R&D company, Go Beyond Consulting Inc., which aims to commercialize the technology for application in environmental monitoring.  The idea is that repeated environmental monitoring surveys could be significantly augmented through the speedy and cost-effective transformation of collected seabed imagery into maps of benthic species.  This would allow timely environmental analysis of changes in species abundance and distribution between successive surveys.  The overall project is known as iSEA:

“iSEA (independent submersible environment analysis) is an automated marine environmental monitoring system.  It will incorporate advanced subsea robotic and camera technology to detect changes in the seabed resulting from natural and/or human activity.  Its purpose is to help protect the world's oceans.

Document Actions