NeMO-Net: A Virtual Expedition

“Science-games” sounds like an oxymoron, doesn’t it? NASA welcomes all gamers and citizen scientists to join them on a virtual expedition to the ocean floor and in due course understand the various ecosystems. This mapping technology can help with studying the health conditions of the coral reefs and help in conservation of the dangerously depleting vibrant coral ecosystems. “NeMO-Net” is a single player game mainly for the iPad. It lets the player classify various ecosystems like corals, algae and sea plants by 3D and 2D painting. Additionally, the players can rate other players, level up and widen their food chain, classify coral reefs and explore the depths from any corner of the world while virtually traveling the ocean in their own research vessel-the “Nautilus”. This game emerged as a result of the publication, “Next-Generation Optical Sensing Technologies for Exploring Ocean Worlds-NASA FluidCam, MiDAR, and NeMO-Net” by Ved Chirayath and Alan Li of NASA Ames Laboratory for Advanced Sensing, United States in the “Frontiers of Marine Science” September 2019 edition.

Data from the “NeMO-Net” game is fed to “NASA NeMO-Net”, the first neural multi-modal observation and training network for global coral reef assessment. It is an open source deep convolutional neural network (CNN) that supports NASA’s Supercomputer-“Pleiades” in classifying the coral reefs and examining their health worldwide. The game comprises various new generation technologies of optical sensing as mentioned above. NeMO-Net exploits active learning and data fusion of millimeter scale remotely sensed 3D images of coral reefs captured using fluid lensing with the NASA FluidCam instrument. Fluid lensing is the phenomenon of interaction of visible light with the aquatic surface waves that develop optical aberrations and form bands of light on the seafloor. This produces refractive lensing that magnifies and demagnifies underwater objects. General Fluid Lensing Algorithm is a novel high-resolution aquatic remote sensing technique for imaging through ocean waves by manipulating the fluid lensing phenomenon. FluidCam is a custom-designed integrated optical system backed with high performance computational instruments. This is the highest-resolution remote sensing benthic imaging technology capable of removing ocean wave distortion. These data are used to train low resolution data from NASA’s Earth Observing System, including hyper spectral airborne remote sensing data and satellite data to globally determine the coral reef ecosystem at unprecedented spatial and temporal scale.

There are various challenges associated with this, such as misrepresentation of coral reef data due to low resolution of remote sensing because of refractive distortion. The NASA FluidCam millimeter scale 3D data shows that the characterization of coral reefs and their related areas are poorly done by current satellite and airborne remote sensing techniques. The current morphological classification is based on kilometer-scale satellites and hence it is highly prone to segmentation errors.

This high accuracy level sensing of 3D imaging can be manipulated to find presence of life elsewhere. The efforts to map stromatolites will help train “NeMO-Net” with the structure and patterns of “biogenic carbonates”. This can then be channeled to make observations in Mars. This technology is a promising step towards grasping the living world outside Earth.

Written by Harshada H
MSc Science and Technology Communication
CSIR-NISCAIR

Instagram link: https://instagram.com/episteme2020?igshid=14o31j9kxmzmt
Mailing address: hharshada934@gmail.com

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: