Skip to content Skip to footer
Deep Vision Logo
Deep Vision Logo
Close
DEEP VISION

Mapping the Unknown Ocean

Harnessing AI to unlock thousands of hours of seafloor imagery and map vulnerable marine ecosystems of the deep Atlantic. Transforming marine conservation through cutting-edge Vision Transformers and machine learning.

Explore the challenge
See Our Impact
0
Camera Deployments
0
Hours of Video
0
Atlantic Nations
0%
Of Planet's Surface
The Challenge

The Deep Ocean Crisis

The deep sea represents 70% of our planet’s surface area and remains largely unexplored. Yet we are increasingly using, and so impacting on this biome.

Certain ecosystems within this vast realm are considered highly vulnerable to human activities, so called Vulnerable Marine Ecosystems (VME). Limited knowledge of where VME are found manifests in stifled and poor decision-making.

Currently, we protect seafloor areas where we have physically observed VME to occur. Meanwhile, we actively license activities in areas for which we have no observations.

The result: poor MPA network design and continuing biodiversity loss.
Our Approach

Modern AI Meets Marine Science

We're using Vision Transformers trained with self-supervised learning to rapidly and accurately extract biodiversity data from an archive of over 4,000 camera platform deployments across the Atlantic Basin.

Vision Transformers

Advanced AI models trained on vast seafloor image libraries using self-supervised learning and fine-tuned with quality-controlled human annotated data to identify VME indicator species with unprecedented accuracy and speed.

Biodiversity Mapping

High-resolution distribution maps of vulnerable marine ecosystems throughout the deep Atlantic Ocean, filling vast data gaps that hinder effective conservation across 8 Atlantic nations.
Expected Impact

Transforming Ocean Conservation

Real-World Results

Our preliminary research applying these methods in UK and Ireland waters identified the extent and distribution of cold-water coral reefs in these nations’ waters, supporting a bottom trawling ban to protect these valuable habitats.

  • Reveal new sites of VME in the deep Atlantic
  • Guide designation of Marine Protected Areas
  • Provide evidence for environmental impact assessments
  • Support regulation of deep-sea mining and fishing activities
  • Enable sustainable ocean management
  • Improve conservation status of deep-sea biodiversity

By involving parties responsible for governing the use of marine space in developing the AI tools and maps, we will facilitate translation of outputs into tangible improvements in conservation policy and action.

— Prof. Kerry Howell, Project Lead

Our Team

Leading Experts in AI & Marine Science

World-class researchers combining deep-sea ecology expertise with cutting-edge artificial intelligence to unlock the secrets of the Atlantic Ocean floor.

Professor Kerry Howell

Professor Kerry Howell

Project Lead
Dr David Moffat

Dr. David Moffat

Lead AI & Machine Learning Scientist
Global Collaboration

Our Partners

Partnering with leading institutions across the Atlantic Basin and a UN Ocean Decade programme to achieve transformative impact.

University of Plymouth
Plymouth Marine Laboratory
University of Bergen
IMR Norway
Galway University
University of Aveiro
University of Beilefeld
University of Gibraltar
UN Ocean Decade Challenger
Proudly funded by

Bezos Earth Fund AI for Climate and Nature Grand Challenge

DeepVision is a proud recipient of seed funding from the Bezos Earth Fund's AI for Climate and Nature Grand Challenge, a global initiative exploring how modern artificial intelligence can multiply the impact of climate and nature conservation efforts.