Broadscale reconnaissance of coral reefs from citizen science and deep learning

Christopher L Lawson,Kathryn M Chartrand,Chris M Roelfsema,Aruna Kolluru,Peter J Mumby

DOI: https://doi.org/10.1101/2024.11.27.625580

2024-12-03

Abstract:Coral reef managers require various forms of data. While monitoring is typically the preserve of scientists, larger scale reconnaissance data that can be used to inform spatial decisions does not usually require such precise measurement. There is an increasing need to collect such broadscale, up to date environmental data at massive scale to prioritise limited conservation resources in the face of global disturbances. Citizen science combined with novel technology presents an opportunity to achieve data collection at the required scale, but the accuracy and feasibility of new tools must be assessed. Here we show that a citizen science program that collects seascape images and analyses them using a combination of deep learning and online citizen scientists can produce accurate benthic cover estimates of key coral groups. The deep learning and citizen scientist analysis methods had different but complementary strengths depending on coral category. When the best performing analysis method was used for each category in all images, mean estimates from 8086 images of percent benthic cover of branching Acropora, plating Acropora, and massive-form coral were ~99% accurate compared to expert assessment of the same images, and >95% accurate at all coral cover ranges tested. The effort to achieve 95% accuracy at a site, our ecologically relevant target based on the accuracy of other tools, was attainable based on citizen scientist involvement in pilot years of the program, with 18-80 images needed depending on coral type and reef state. Power analyses showed that sampling up to 114 images per site was needed to detect a 10% absolute difference in coral cover per category (power = 0.8), accounting for natural heterogeneity. However, the benthic cover of all other coral groups as a single category could only be estimated with 95% accuracy at 60% of survey sites and for images with 10-30% coral cover. Disaggregating this other coral group into more distinct coral categories may improve accuracy. Overall, citizen science can provide an accuracy that is acceptable for many end-users for select coral morphologies. Such a combination of emerging technology and citizen science presents an attainable tool for collecting inexpensive, widespread reconnaissance data of coral reefs that can complement higher resolution survey programs or be an accessible tool for resource-poor locations.

Ecology

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to evaluate whether the coral reef landscape images collected by citizen science projects can provide reliable enough information to support coral reef management decisions. Specifically, the research objectives include: 1. **Evaluating the accuracy of AI - independent analysis and AI - assisted citizen scientist analysis**: The research aims to verify whether these two methods can accurately quantify the coverage rates of three key coral communities (branching Acropora, tabular Acropora, and massive corals) in the coral reef landscape images collected by citizen scientists. 2. **Determining the number of images required to reach a reasonable level of accuracy**: Given the differences in accuracy among different images, the research explores how many images are needed to reach a reasonable level of accuracy at a survey site, and how many online citizen scientists are required to analyze each image. 3. **Conducting a power analysis**: Through a series of power analyses, the research determines how many images need to be collected from a site to detect an absolute difference of 10% for each coral category (with a power of 0.8), while taking into account the heterogeneity of the natural environment. The core of these objectives lies in exploring how to combine emerging technologies (such as deep learning) and citizen science to achieve large - scale, low - cost collection of coral reef environmental data, in order to supplement high - resolution monitoring projects or serve as a viable tool in resource - poor areas.

Broadscale reconnaissance of coral reefs from citizen science and deep learning

Digitizing the coral reef: machine learning of underwater spectral images enables dense taxonomic mapping of benthic habitats

Monitoring of Coral Reefs Using Artificial Intelligence: A Feasible and Cost-Effective Approach

Using Remote Sensing and in situ Measurements for Efficient Mapping and Optimal Sampling of Coral Reefs

A benthic substrate classification method for seabed images using deep learning: application to management of deep‐sea coral reefs

Combining Drones and Deep Learning to Automate Coral Reef Assessment with RGB Imagery

Scalable semantic 3D mapping of coral reefs with deep learning

A Citizen Science Approach: A Detailed Ecological Assessment of Subtropical Reefs at Point Lookout, Australia

Mapping the world's coral reefs using a global multiscale earth observation framework

Deep Learning-Based Automatic Estimation of Live Coral Cover from Underwater Video for Coral Reef Health Monitoring

Improving coral monitoring by reducing variability and bias in cover estimates from seabed images

Monitoring through many eyes: Integrating disparate datasets to improve monitoring of the Great Barrier Reef

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

Increasing trust in new data sources: crowdsourcing image classification for ecology

An evaluation of the error and uncertainty in epibenthos cover estimates from AUV images collected with an efficient, spatially-balanced design

Confidence Levels, Sensitivity, and the Role of Bathymetry in Coral Reef Remote Sensing

Application of Photogrammetric Computer Vision and Deep Learning in High-Resolution Underwater Mapping: A Case Study of Shallow-Water Coral Reefs

Assessing the Potential of Remotely-Sensed Drone Spectroscopy to Determine Live Coral Cover on Heron Reef

Mapping emergent coral reefs: a comparison of pixel‐ and object‐based methods

CoralSCOP-LAT: Labeling and Analyzing Tool for Coral Reef Images with Dense Mask