Improved PSP and U-Net Architectures for Forest Segmentation in Remote Sensing Pictures
Authors:
- 14 November 2022
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 614-618
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 A veragePooling2D, 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.Keywords
This publication has 12 references indexed in Scilit:
- Forest and Water Bodies Segmentation Through Satellite Images Using U-NetESS Open Archive, 2022
- Semantic Segmentation of Satellite Images Using Deep-UnetArabian Journal for Science and Engineering, 2022
- Improving a neural network model for semantic segmentation of images of monitored objects in aerial photographsEastern-European Journal of Enterprise Technologies, 2021
- Improvement of the model of object recognition in aero photographs using deep convolutional neural networksEastern-European Journal of Enterprise Technologies, 2021
- Development of brain tumor segmentation of magnetic resonance imaging (MRI) using U-Net deep learningEastern-European Journal of Enterprise Technologies, 2021
- Construction of an advanced method for recognizing monitored objects by a convolutional neural network using a discrete wavelet transformEastern-European Journal of Enterprise Technologies, 2021
- Extending deep learning approaches for forest disturbance segmentation on very high‐resolution satellite imagesRemote Sensing in Ecology and Conservation, 2021
- Semantic Image Segmentation for Building Detection in Urban Area with Aerial Photograph Image using U-Net Models2020
- Superpixel Segmentation With Fully Convolutional Networks2020
- M-UNet: Modified U-Net Segmentation Framework with Satellite Imagery2020