Descripción
|
|
---|---|
Abstract. Infrastructure detection and monitoring traditionally required manual identification of geospatial objects in aerial imagery but advances in deep learning and computer vision enabled the researchers in the field of remote sensing to successfully apply transfer learning from pretrained models on largescale datasets for the task of geospatial object detection. However, they mostly focused on objects with clearly defined boundaries that are independent of the background (e.g. airports, airplanes, buildings, ships, etc.). What happens when we have to deal with more complicated, continuous objects like roads? In this paper we will review four of the best-known CNN architectures (VGGNet, Inception-V3, Xception, Inception-ResNet) and apply feature extraction and fine-tuning techniques to detect the existence of roads in aerial orthoimages divided in tiles of 256 ? 256 pixels in size. We will evaluate each model¿s performance on unseen test data using the accuracy metric and compare the results with those obtained by a CNN especially built for this purpose. Keywords: Transfer learning Convolutional neural networks Remote sensing Road detection | |
Internacional
|
Si |
Nombre congreso
|
Second International Conference, ICAI 2019, |
Tipo de participación
|
960 |
Lugar del congreso
|
Madrid |
Revisores
|
Si |
ISBN o ISSN
|
978-3-030-32474-2 |
DOI
|
https://doi.org/10.1007/978-3-030-32475-9_14 |
Fecha inicio congreso
|
06/11/2019 |
Fecha fin congreso
|
08/11/2019 |
Desde la página
|
185 |
Hasta la página
|
198 |
Título de las actas
|
Communications in Computer and Information Science, vol 1051. Springer, Cham |