Improved PSP and U-Net architectures for forest segmentation in remote sensing pictures

Loading...
Thumbnail Image
Date
2022-11-14
Authors
Slyusar, V. I.
Sliusar, I. I.
Pavlenko, A.
Journal Title
Journal ISSN
Volume Title
Publisher
IEEE
Abstract
Various PSP and U-Net architectures for forest segmentation in remote sensing pictures have been proposed and investigated. The main improvements of the proposed architectures are based on using the BathNormalization layers, replacing MaxPool2D layers with AveragePooling2D, changing Conv2DTranspose to UpSampling2D blocks, etc. For the training of neural networks was used modified dataset of 128x128 pictures based on the dataset from Kaggle. As a result of improving architecture was given the maximum segmentation accuracy of 80.8 % on the validation set of pictures.
Description
Slyusar V., Sliusar I., Pavlenko A. Improved PSP and U-Net architectures for forest segmentation in remote sensing pictures.// IEEE 2nd Ukrainian Microwave Week (Virtual Event) IEEE UkrMW-2022, 14 – 18 November, 2022. - V. N. Karazin Kharkiv National University, Kharkiv, Ukraine. – 4 p. DOI: 10.1109/UkrMW58013.2022.10037105.
Keywords
Citation