Implementing Convolutional Neural Networks and Vision Transformers for Satellite Image Processing [Articol]
Date
2024
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This paper explores the application of Vision Transformers (ViT) for satellite imagery classification, inspired by the widespread use of transformers in language processing and related computer vision research. The study involved training a ViT model on the EuroSAT dataset and comparing its performance with ResNet50. Additionally, a hybrid model architecture integrating CNNs with ViTs was proposed to leverage the strengths of both approaches.
Description
Keywords
visual transformers, convolutional neural networks, satellite image, datasets, image recognition
Citation
ȚURCAN, Matei-Octavian and Olesea CAFTANATOV. Implementing Convolutional Neural Networks and Vision Transformers for Satellite Image Processing. In: International Conference dedicated to the 60th anniversary of the foundation of Vladimir Andrunachievici Institute of Mathematics and Computer Science, MSU, October 10-13 2024. Chisinau: [S. n.], 2024, pp. 387-396. ISBN 978-9975-68-515-3.