Overview

The goal of this report is to identify mobile fauna activity around artificial and natural reefs in the Maldives.

For this project, seven underwater camera systems - Unbaited Remote Underwater Video Systems (URUVS), were deployed to collect high-resolution footage of marine life in a non-intrusive way. The data was processed using our AnemoAI model, trained on around 3,000 annotated frames. This allowed for the identification and classification of species observed in the footage, providing an efficient method for analyzing marine biodiversity.

Process

1. Project Specification
2. Data Collection & Video Recording
3. Data Annotation & AI Training
4. AI Analysis
5. Data Visualization

Manual Labeling Findings

• Larger fish species (such as sea cub, surgeonfish, sharks, and sweetlips) are predominantly observed around the older, rocky, and more open natural reef. In contrast, smaller fish species (including damselfish, angelfish, and butterflyfish) are primarily observed around the artificial reef structures.
• Schools of juvenile fish are observed exclusively on the artificial reefs, further supporting the hypothesis that these structures provide important refuge and nursery habitats.
• Fish on the artificial reefs frequently interact with the structure itself, swimming through crevices and remaining close to it, whereas fish on the natural reef are more widely dispersed.