Coral Detection, Ranging, and Assessment (CDRA) algorithm-based automatic estimation of coral reef coverage

Abstract

Climate change poses significant threats to the health of coral reefs. The detection of coral coverage is a quantitative method for assessing the health of coral reefs. Valuable insights into the effects of climate change on coral reef ecosystems and the trends in coral reef development can be gained through the monitoring of coral coverage. However, traditional manual methods for estimating coral coverage are time-consuming and labor-intensive, limiting the scalability and efficiency of coral surveys. The field of automated processing for coral video imagery is currently limited, impeding the development of effective techniques for comprehensive coral monitoring and further hindering progress in coral ecosystem monitoring. To cope with these challenges, there is a crucial need for automated systems to efficiently handle large volumes of video data, thereby transforming imaging devices, such as underwater robots, into autonomous sensors capable of conducting reliable and comprehensive surveys. In response to this pressing issue, this study presents a novel video analysis approach, termed the Coral Detection, Ranging, and Assessment (CDRA) algorithm, which combines cutting-edge computer vision techniques including You Only Look Once version 5 (YOLOv5) object detection, binocular stereo vision, and tracking algorithms. The CDRA algorithm aims to analyze continuous video segments of coral reefs, enabling the identification, counting, and estimation of size and location of individual coral colonies. Experimental evaluations conducted in controlled pool environments demonstrated the high accuracy of the proposed method, with an average accuracy of 90.7% in estimating coral coverage at different locations. Moreover, field trials conducted in Xidao Island of China verified the effectiveness and robustness of the CDRA algorithm under varying water quality and lighting conditions. The findings of this study represent a significant step towards the development of reliable and automated techniques for coral reef detection, contributing to the advancement of coral reef conservation efforts.