Chirayath, Ved: University of Miami (Project Lead)
Project Funding:
2016 - 2019
NRA: 2016 NASA: Advanced Information Systems Technology
Funded by NASA, Other US Funding:
Abstract:
We propose NeMO-Net, the first neural multi-modal observation and training network for global coral reef assessment. NeMO-Net is an open-source deep convolutional neural network (CNN) and interactive active learning training software proposed to accurately assess the present and past dynamics of coral reef ecosystems through determination of percent living cover and morphology as well as mapping of spatial distribution. NeMO-Net exploits active learning and data fusion of mm-scale remotely sensed 3D images of coral reefs from our ESTO FluidCam instrument, presently the highest-resolution remote sensing benthic imaging technology capable of removing ocean wave distortion, as well as lower-resolution airborne remote sensing data from the ongoing NASA CORAL mission and satellite data to determine coral reef ecosystem makeup globally at unprecedented spatial and temporal scales.Aquatic ecosystems, particularly coral reefs, remain quantitatively misrepresented by low-resolution remote sensing as a result of refractive distortion from ocean waves, optical attenuation, and remoteness. ESTO-funded machine learning classification of coral reefs using FluidCam mm-scale 3D data showed that present satellite and airborne remote sensing techniques poorly characterize fundamental coral reef health indicators, such as percent living cover, morphology type, and species breakdown at the cm and meter scale. Indeed, current global assessments of coral reef cover and morphology classification based on km-scale satellite data alone can suffer from segmentation errors greater than 40%, capable of change detection only on yearly temporal scales and decameter spatial scales, significantly hindering our understanding of patterns and processes in marine biodiversity at a time when these ecosystems are experiencing unprecedented anthropogenic pressures, ocean acidification, and sea surface temperature rise.NeMO-Net leverages ESTO investment in our augmented machine learning algorithm that demonstrates data fusion of regional FluidCam (mm, cm-scale) airborne remote sensing with global low-resolution (m, km-scale) airborne and spaceborne imagery to reduce classification errors up to 80% over regional scales. Such technologies can substantially enhance our ability to assess coral reef ecosystems dynamics using NASA EOS data. Through unique international partnerships with the IUCN Global Marine Program, Dr. Sylvia Earle’s Mission Blue, NASA’s CORAL, HICE-PR, and CoralBASICS projects, we are working directly with target recipient communities on NeMO-Net to train the largest aquatic neural network and produce an impactful technology development that has real-world scientific and policy impacts.Our project goals are to: (1) create a fused global dataset of coral reefs from FluidCam, CORAL, and NASA EOS data, (2) train NeMO-Net’s CNN through active learning via an interactive app and global partners, (3) develop the NeMO-Net CNN architecture, (4) perform global coral reef assessment using NeMO-Net and determine the spatial distribution, percent living cover, and morphology breakdown of corals at present and over the past decade at meter spatial scales and weekly intervals, (5) evaluate NeMO-Net CNN error and robustness against existing unfused methods and (6) deploy NeMO-Net as a NASA NEX and QGIS open-source module for use in the community.NeMO-Net is relevant to AIST Data-Centric Technologies in data fusion and data mining, as well as special subtopics subsection 3.2.1 (a-d), (f-i) through autonomous integration of data from sensors of various observational capacities to form data products across a wide range of spatial and temporal domains. NeMO-Net has broad applicability to data fusion for automated assessment of both terrestrial and aquatic ecosystems. The period of performance for this project spans 2 years and leverages significant previous ESTO investments in our technology for an entry TRL of 2 and exit of TRL of 4.
Publications:
Chirayath, V., Li, A. 2019. Next-Generation Optical Sensing Technologies for Exploring Ocean Worlds--NASA FluidCam, MiDAR, and NeMO-Net. Frontiers in Marine Science. 6. DOI: 10.3389/fmars.2019.00521
More details may be found in the following project profile(s):
