Computer Vision Coral Ecology: Applying Automated Computer Vision Methods for Marine Conservation
(2/4/2014) 9 minutes

description:

Coral reefs are essential to coastal societies throughout the world, providing food, resources and income to over 500 million people. In the last three decades up to 80% of Caribbean coral coverage has been lost and up to 50% in the Indo-Pacific. A Scripps team of biologists and engineers led by Greg Mitchell, David Kline and Jules Jaffe have teamed up with a UCSD Computer Vision team led by David Kreigman and Serge Belongie to develop an automated computer learning system to increase the speed and coverage of coral reef surveying 100-1000x. A website, coralnet.ucsd.edu, has been developed where ~30 coral reef research groups have uploaded over 30,000 reef photos and provided over 1 million manual annotations. CoralNet can provide automated annotation on new images at around 80% accuracy compared to expert coral reef ecologists.

more on this subject: