Scalable Semantic 3D Mapping of Coral Reefs with Deep Learning

TL;DR

This paper introduces a scalable, automated system for high-resolution 3D mapping and semantic analysis of coral reefs using deep learning, significantly reducing manual effort and costs, and enabling rapid environmental monitoring.

Contribution

It presents a novel integrated approach combining machine learning-based 3D mapping with semantic segmentation for underwater environments, enhancing scalability and automation.

Findings

High-precision 3D semantic mapping achieved at scale

Analysis of a 100 m video transect within 5 minutes

Significant reduction in labor and equipment costs

Abstract

Coral reefs are among the most diverse ecosystems on our planet, and are depended on by hundreds of millions of people. Unfortunately, most coral reefs are existentially threatened by global climate change and local anthropogenic pressures. To better understand the dynamics underlying deterioration of reefs, monitoring at high spatial and temporal resolution is key. However, conventional monitoring methods for quantifying coral cover and species abundance are limited in scale due to the extensive manual labor required. Although computer vision tools have been employed to aid in this process, in particular SfM photogrammetry for 3D mapping and deep neural networks for image segmentation, analysis of the data products creates a bottleneck, effectively limiting their scalability. This paper presents a new paradigm for mapping underwater environments from ego-motion video, unifying 3D mapping systems that use machine learning to adapt to challenging conditions under water, combined with a modern approach for semantic segmentation of images. The method is exemplified on coral reefs in the northern Gulf of Aqaba, Red Sea, demonstrating high-precision 3D semantic mapping at unprecedented scale with significantly reduced required labor costs: a 100 m video transect acquired within 5 minutes of diving with a cheap consumer-grade camera can be fully automatically analyzed within 5 minutes. Our approach significantly scales up coral reef monitoring by taking a leap towards fully automatic analysis of video transects. The method democratizes coral reef transects by reducing the labor, equipment, logistics, and computing cost. This can help to inform conservation policies more efficiently. The underlying computational method of learning-based Structure-from-Motion has broad implications for fast low-cost mapping of underwater environments other than coral reefs.