EESS talk - Student presentation on "Scalable 3D Semantic Mapping of Coral Reefs with Deep Learning"
Abstract:
Coral reefs, which host more than a third of the ocean’s biodiversity on less than 0.1% of its surface, are existentially threatened by climate change and other human activities. This necessitates methods for evaluating the state of coral reefs that are efficient, scalable, and low-cost. Current digital reef monitoring tools typically rely on conventional Structure-from-Motion photogrammetry, which can limit the scalability, and current datasets for training semantic segmentation systems are either sparsely labeled, domain-specific, or very small. We describe how deep-learning-based 3D semantic mapping can enable rapid mapping of coral reef transects, even when using low-cost underwater cameras. The 3D mapping component learns to tackle the challenging lighting effects of underwater environments from a large dataset of reef videos. The transnational data-collection initiative was carried out in Djibouti, Sudan, Jordan, and Israel. This research paves the way for affordable and widespread deployment of the method in analysis of video transects in conservation and ecology, highlighting a promising intersection with machine learning for tangible impact in understanding these oceanic ecosystems.
Biography:
Jonathan Sauder is currently a PhD student at the Environmental Computational Sciences and Earth Observation (ECEO) Laboratory and the Laboratory for Biological Geochemistry (LGB) at EPFL, and is part of the Transnational Red Sea Center. His research focuses on developing the next generation of coral reef monitoring tools driven by deep learning. Previously, he completed his Bachelor’s at the Hasso Plattner Institute at the University of Potsdam and his Master’s at TU Berlin, both in computer science.
Coral reefs, which host more than a third of the ocean’s biodiversity on less than 0.1% of its surface, are existentially threatened by climate change and other human activities. This necessitates methods for evaluating the state of coral reefs that are efficient, scalable, and low-cost. Current digital reef monitoring tools typically rely on conventional Structure-from-Motion photogrammetry, which can limit the scalability, and current datasets for training semantic segmentation systems are either sparsely labeled, domain-specific, or very small. We describe how deep-learning-based 3D semantic mapping can enable rapid mapping of coral reef transects, even when using low-cost underwater cameras. The 3D mapping component learns to tackle the challenging lighting effects of underwater environments from a large dataset of reef videos. The transnational data-collection initiative was carried out in Djibouti, Sudan, Jordan, and Israel. This research paves the way for affordable and widespread deployment of the method in analysis of video transects in conservation and ecology, highlighting a promising intersection with machine learning for tangible impact in understanding these oceanic ecosystems.
Biography:
Jonathan Sauder is currently a PhD student at the Environmental Computational Sciences and Earth Observation (ECEO) Laboratory and the Laboratory for Biological Geochemistry (LGB) at EPFL, and is part of the Transnational Red Sea Center. His research focuses on developing the next generation of coral reef monitoring tools driven by deep learning. Previously, he completed his Bachelor’s at the Hasso Plattner Institute at the University of Potsdam and his Master’s at TU Berlin, both in computer science.
Practical information
- General public
- Free
- This event is internal
Organizer
- EESS - IIE
Contact
- Prof.Devis Tuia, ECEO