No abstract

No abstract

No abstract

There is a severe demand for, and shortage of, large accurately labeled datasets to train supervised computational intelligence (CI) algorithms in domains like unmanned aerial systems (UAS) and autonomous vehicles. This has hindered our ability to develop and deploy various computer vision algorithms in/across environments and niche domains for tasks like detection, localization, and tracking. Herein, I propose a new human-in-the-loop (HITL) based growing neural gas (GNG) algorithm to minimize human intervention during labeling large UAS data collections over a shared geospatial area. Specifically, I address human driven events like new class identification and mistake correction. I also address algorithm-centric operations like new pattern discovery and self-supervised labeling. Pattern discovery and identification through self-supervised labeling is made possible through open set recognition (OSR). Herein, I propose a classifier with the ability to say "I don't know" to identify outliers in the data and bootstrap deep learning (DL) models, specifically convolutional neural networks (CNNs), with the ability to classify on N+1 classes. The effectiveness of the algorithms are demonstrated using simulated realistic ray-traced low altitude UAS data from the Unreal Engine. The results show that it is possible to increase speed and reduce mental fatigue over hand labeling large image datasets.