Using Deep Learning to Monitor Coral Reef Health
By Rithika Narayan
A few years ago, my best friend returned from her vacation to the island of Cozumel off the coast of Mexico with a GoPro full of images she had taken while paragliding, hiking, and relaxing on the beach. Most interesting to me, however, were the photos she had taken while scuba diving in the coral reefs that surround the island. As we scrolled through dozens of pictures of corals, I realized that my friend, like millions of other tourists, had in her possession a treasure trove of information about the health of the coral reefs of Cozumel at the time she visited. However, the issue was how to collect and analyze all that information efficiently … Machine learning (ML) is a field of computer science concerning the study of algorithms that learn from exposure to previous data in order to improve at autonomously completing a task. Chances are you’ve already interacted with ML algorithms in your everyday life: they drive the show recommendations that Netflix gives you, virtual assistants like Siri and Alexa, and facial recognition software. I developed my interest in this field as I was thinking about my future in college and beyond; in fact, this project was a principal factor behind my decision to study computer science in college. I decided to dive head first into exploring ML by applying it environmental monitoring. With the guidance of my mentor Mr. Anthony Pellicano, the ML specialist at Angion Biomedica Corp. in New York, I used a convolutional neural network to analyze underwater images of coral reefs in order to detect the presence of healthy and bleached corals within these images.