Autonomous underwater vehicles (AUVs) equipped
with in-situ artificial intelligence (AI)-enabled image
processing capability could help understand the influence
of climate change on reef ecology, and evolve wellinformed policy decisions for sustaining coral reefs.
The recent focus is on developing technologies to collect
images of coral reefs autonomously over larger areas
using AUVs with AI-aided image processing capabilities.
This article presents the development of a computer
vision and deep-learning-based coral identification and
tracking algorithm trained on six distinct families of
coral images acquired from the Andaman Islands using a
remotely operated vehicle PROVe500 developed by the
National Institute of Ocean Technology, Chennai. The
YOLOv8-trained coral detection and classification
model, along with the ByteTrack, have been used to
identify and track the corals with mean average precision
and recall rates of 97.4% and 96.8% respectively. The
developed AI algorithm gives confidence in deploying
AUVs to identify and track corals of interest or anomalies in real-time over a large spatial domain.