High-Level Project Summary
We developed two AI methods to detect and track the plastic debris in the ocean. One method is anomaly detection according to ocean surface wind speed. (Wind speed is different where the debris are.) The second method is clustering the reflection values of ocean surface radar signals. (Reflectability and velocity of the signals are different on water and debris. ) We also developed an interactive dashboard to put the result of our analysis and visualize it in an attractive easy-to-understand way. The user can choose the region of interest in the ocean map and also choose a timestamp, and select the preferred detection method, and regarding those inputs, the result will show up as plots.
Link to Project "Demo"
Link to Final Project
Detailed Project Description
PosAIdon aims to provide accurate analysis of global marine debris. An interactive console will enable the users to see in which areas of the sea marine debris is accumulated. While we need global efforts to curb marine debris, it is also difficult to build any inter country collaboration. Using our console activists, government organizations and green entrepreneurs will be able to select area under their jurisprudence or access and take action based on our analysis visualized by multiple python frameworks and created by our propreitory Artificial Intelligence algorithm. We aim to provide accurate predictions by basing our algorithm on multiple features present in available datasets like wind speed and radar reflectance of sea. Our algorithm would not require any more collection of data.
In our interactive console, users can simply use click and drag to see any possible debris in the area. Our product does not just stand out for its integration of multiple datasets for better predictions but also for its simplicity and usability.
Space Agency Data
We used all data from NASA earth data. We were inspired by a study conducted by Evans et al. [1], which uses the difference between measured and predicted ocean surface roughness for a given wind speed as an indicator for the presence of Micro-plastic concentration in the Ocean. We aimed to extend this idea to include other data sources further to have more confidence in predictions. Reflection, absorption, and transmittance of electromagnetic radiation are highly dependent on the concentrations, types, and presence of substances in water.
We have a working algorithm for anomaly detection algorithm introduced by Evans et al. [1]. We also aim to extend the algorithm to use Terra Ocean Reflectance datasets from the NASA Earth Data repository to increase our anomaly detection accuracy.
[1] Evans, M. and Ruf, C., 2021. Toward the Detection and Imaging of Ocean Micro-plastics With a Spaceborne Radar. IEEE Transactions on Geoscience and Remote Sensing, pp.1-9.
Hackathon Journey
since we have studied masters in computer science and our specialization is AI, it was very exciting for us to take part in AI challenge of NASA Hackathon to get to know some real-world AI problems. It was also very challenging and rewarding to think about the problems, come with the solution, try different methods, read lots of articles, and implement our idea.
Also working as a team was really beautiful and a great challenge to handle especially when 2 of us already knew each other and one was new. And the 2 were working in presence and the one in Home Office. So we learned a lot about the team work.
Since I (Maryam Arabshahi) am an environment activist and also an AI lover, it is my dream to use AI to help the environment and for SDG 17 (Sustainable Development Goals) projects. So, it was an honor for me to work on this challenge and I am very happy about it.
References
Libraries:
Basemap
Matplotlib
math
netCDF4
folium
ipywidgets
IPython
Programming Language: Python
Tools:
Jupyter Notebook
Atom IDE
Tags
#AI, #Debris, #Artificial_Intelligence, #Anomaly_Detection, #Clustering, #MicroPlastics, #Oceans
Global Judging
This project has been submitted for consideration during the Judging process.