Using Neural Networks and Image Labeling for Artificial Reef Analysis
School Name
South Carolina Governor's School for Science & Mathematics
Grade Level
12th Grade
Presentation Topic
Computer Science
Presentation Type
Mentored
Abstract
Our research was centered around creating training data for two neural networks that would allow aquatic robots to move autonomously and identify coral using a camera. Training data was created using images taken from previous expeditions to artificial coral reefs created from sunken ships. We focused on creating accurate training data for use in the python based neural networks that the other students were tasked with creating. By the end of our research term, we had created a large enough quantity of training data for the robot to correctly identify test images of coral for the most part. In the future, even more training data will be created and the neural networks will be fine-tuned and tested until these robots are ready to deploy for data collection of coral reefs across the world.
Recommended Citation
Kay, Asher and Sethi, Rajat, "Using Neural Networks and Image Labeling for Artificial Reef Analysis" (2020). South Carolina Junior Academy of Science. 148.
https://scholarexchange.furman.edu/scjas/2020/all/148
Location
Furman Hall 109
Start Date
3-28-2020 10:45 AM
Presentation Format
Oral Only
Group Project
Yes
Using Neural Networks and Image Labeling for Artificial Reef Analysis
Furman Hall 109
Our research was centered around creating training data for two neural networks that would allow aquatic robots to move autonomously and identify coral using a camera. Training data was created using images taken from previous expeditions to artificial coral reefs created from sunken ships. We focused on creating accurate training data for use in the python based neural networks that the other students were tasked with creating. By the end of our research term, we had created a large enough quantity of training data for the robot to correctly identify test images of coral for the most part. In the future, even more training data will be created and the neural networks will be fine-tuned and tested until these robots are ready to deploy for data collection of coral reefs across the world.