The Impact of Neural Network Complexity on the Detection of Rockfalls on Mars and the Moon
Scholarly Review Journal · 2022
DOI: 10.70121/001c.121657 ↗Kurz gefasst
Fachartikel in der Zeitschrift Scholarly Review Journal, 2022. Verfasst von Henry Zhao. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The detection of rockfalls is important to real-time tasks on celestial objects, such as finding potentially hazardous or geologically active areas on Mars. Currently, rockfall processing is done by hand, requiring a large amount of human labor. This study intends to test the effectiveness of machine learning on the detection of rockfalls on Mars and the Moon. Specifically, this study tests the impact of neural network complexity, and thus detection time, on the precision, recall, and F1 score of detection results. 5 different convolutional neural networks were trained and tested on a dataset of Mars and moon rockfalls. It was found that Resnet50 with transfer learning performed the best, with a median detection time and recall, and high precision and F1. There also is a rough pattern corresponding longer detection times with more precision and less recall, and shorter detection time with more recall and less precision. From the data gathered, it appears that it is feasible to use machine learning to quickly and effectively detect rockfalls on Mars, for use in real-time processing and decision making.
Zitieren
APA
Zhao, H. (2022). The Impact of Neural Network Complexity on the Detection of Rockfalls on Mars and the Moon. Scholarly Review Journal. https://doi.org/10.70121/001c.121657
Vancouver
Zhao H. The Impact of Neural Network Complexity on the Detection of Rockfalls on Mars and the Moon. Scholarly Review Journal. 2022. doi:10.70121/001c.121657
BibTeX
@article{Zhao2022,
author = {Zhao, Henry},
title = {The Impact of Neural Network Complexity on the Detection of Rockfalls on Mars and the Moon},
journal = {Scholarly Review Journal},
year = {2022},
doi = {10.70121/001c.121657}
}
EUVIMED