Training Artificial Intelligence Algorithms with Automatically Labelled UAV Data from Physics-Based Simulation Software

• Published in: Applied sciences Vol. 13; no. 1; p. 131
• Main Authors: Boone, Jonathan, Goodin, Christopher, Dabbiru, Lalitha, Hudson, Christopher, Cagle, Lucas, Carruth, Daniel
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01-01-2023
• Subjects: Algorithms; Artificial intelligence; Automation; Computer simulation; convolutional neural networks; Data acquisition; Datasets; Image acquisition; Learning algorithms; Machine learning; Micro air vehicles (MAV); Neural networks; Sensors; simulator; smart trained models; synthetic image data; Unmanned aerial vehicles; Workflow
• ISSN: 2076-3417; 2076-3417
• DOI: 10.3390/app13010131

Machine-learning (ML) requires human-labeled “truth” data to train and test. Acquiring and labeling this data can often be the most time-consuming and expensive part of developing trained models of convolutional neural networks (CNN). In this work, we show that an automated workflow using automatically labeled synthetic data can be used to drastically reduce the time and effort required to train a machine learning algorithm for detecting buildings in aerial imagery acquired with low-flying unmanned aerial vehicles. The MSU Autonomous Vehicle Simulator (MAVS) was used in this work, and the process for integrating MAVS into an automated workflow is presented in this work, along with results for building detection with real and simulated images.