Development of Coral Investigation System Based on Semantic Segmentation of Single-Channel Images

Among aquatic biota, corals provide shelter with sufficient nutrition to a wide variety of underwater life. However, a severe decline in the coral resources can be noted in the last decades due to global environmental changes causing marine pollution. Hence, it is of paramount importance to develop and deploy swift coral monitoring system to alleviate the destruction of corals. Performing semantic segmentation on underwater images is one of the most efficient methods for automatic investigation of corals. Firstly, to design a coral investigation system, RGB and spectral images of various types of corals in natural and artificial aquatic sites are collected. Based on single-channel images, a convolutional neural network (CNN) model, named DeeperLabC, is employed for the semantic segmentation of corals, which is a concise and modified deeperlab model with encoder-decoder architecture. Using ResNet34 as a skeleton network, the proposed model extracts coral features in the images and performs semantic segmentation. DeeperLabC achieved state-of-the-art coral segmentation with an overall mean intersection over union (IoU) value of 93.90%, and maximum F1-score of 97.10% which surpassed other existing benchmark neural networks for semantic segmentation. The class activation map (CAM) module also proved the excellent performance of the DeeperLabC model in binary classification among coral and non-coral bodies.

A Review of Current Coral Monitoring Tools/ Diagnostic Methods & Introducing a New Tool to the Coral Health Toolkit

Rapidly and repeatedly ascertaining the health status of coral reefs is an ever more pressing issue as part of activities to understand and monitor the damaging impacts of climate change. A combination of increasing ocean temperatures, acidity and frequency of extreme storm events continues to alter the marine environment beyond what sensitive organisms, such as coral, can cope with. It is therefore vital to establish technologies and validated methods to provide a metric or indication into the health of these organisms. There are currently many surveys and techniques used by coral scientists to uncover insights into the status and assessment of coral reefs, from colour wheels to multispectral satellite surveys. Here we outline an array of current techniques and methods focused specifically on coral monitoring and health diagnosis, ranging across the length scales from simple diver-based surveyance to satellite remote sensing. The technique of using hyperspectral fluorescence imaging is also introduced as a viable novel addition to aid and extend the current toolbox of available technologies.

The use of pseudo-multivariate standard error to improve the sampling design of coral monitoring programs

The characteristics of coral reef sampling and monitoring are highly variable, with numbers of units and sampling effort varying from one study to another. Numerous works have been carried out to determine an appropriate effect size through statistical power; however, these were always from a univariate perspective. In this work, we used the pseudo multivariate dissimilarity-based standard error (MultSE) approach to assess the precision of sampling scleractinian coral assemblages in reefs of Venezuela between 2017 and 2018 when using different combinations of number of transects, quadrats and points. For this, the MultSE of 36 sites previously sampled was estimated, using four 30m-transects with 15 photo-quadrats each and 25 random points per quadrat. We obtained that the MultSE was highly variable between sites and is not correlated with the univariate standard error nor with the richness of species. Then, a subset of sites was re-annotated using 100 uniformly distributed points, which allowed the simulation of different numbers of transects per site, quadrats per transect and points per quadrat using resampling techniques. The magnitude of the MultSE stabilized by adding more transects, however, adding more quadrats or points does not improve the estimate. For this case study, the error was reduced by half when using 10 transects, 10 quadrats per transect and 25 points per quadrat. We recommend the use of MultSE in reef monitoring programs, in particular when conducting pilot surveys to optimize the estimation of the community structure.