Extraction of Built-Up Areas Using Convolutional Neural Networks and Transfer Learning From Sentinel-2 Satellite Images

Bramhe, Vijendra Singh; Ghosh, S. K.; Garg, Pradeep

doi:10.5194/isprs-archives-xlii-3-79-2018

Cited by 18 publications

(12 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, little effort has been directed towards the challenge of large-scale built-up areas mapping with CNN from data of lower spatial resolution such as the ones powered by Sentinel-2. The works of [51], [52] represent a significant advancement in that direction. In particular, the framework of human settlements mapping proposed at 20 m by [52] is a step-forward towards a global scale model.…”

Section: Introductionmentioning

confidence: 99%

Convolutional neural networks for global human settlements mapping from Sentinel-2 satellite imagery

Corbane

Syrris

Sabo³

et al. 2020

Neural Comput & Applic

109

View full text Add to dashboard Cite

Spatially consistent and up-to-date maps of human settlements are crucial for addressing policies related to urbanization and sustainability, especially in the era of an increasingly urbanized world. The availability of open and free Sentinel-2 data of the Copernicus Earth Observation program offers a new opportunity for wall-to-wall mapping of human settlements at a global scale. This paper presents a deep-learning-based framework for a fully automated extraction of built-up areas at a spatial resolution of 10 m from a global composite of Sentinel-2 imagery. A multi-neuro modeling methodology building on a simple Convolution Neural Networks architecture for pixel-wise image classification of built-up areas is developed. The core features of the proposed model are the image patch of size 5 × 5 pixels adequate for describing built-up areas from Sentinel-2 imagery and the lightweight topology with a total number of 1,448,578 trainable parameters and 4 2D convolutional layers and 2 flattened layers. The deployment of the model on the global Sentinel-2 image composite provides the most detailed and complete map reporting about built-up areas for reference year 2018. The validation of the results with an independent reference dataset of building footprints covering 277 sites across the world establishes the reliability of the built-up layer produced by the proposed framework and the model robustness. The results of this study contribute to cutting-edge research in the field of automated built-up areas mapping from remote sensing data and establish a new reference layer for the analysis of the spatial distribution of human settlements across the rural–urban continuum.

show abstract

Section: Introductionmentioning

confidence: 99%

Convolutional neural networks for global human settlements mapping from Sentinel-2 satellite imagery

Corbane

Syrris

Sabo³

et al. 2020

Neural Comput & Applic

109

View full text Add to dashboard Cite

show abstract

“…It has also been able to differentiate between the number of built-up areas with a higher correlation than other indices used in urban areas. The three known spectral indices, namely NDBI, BAEI, and NDBaI was utilized for multispectral groups gained via Landsat 8 OLI sensor, Landsat-8 imaginary obtained the built-up indices are explicit to feature built-up class in which useful in planning built-up area by Bramhe (2018).…”

Section: Literature Reviewmentioning

confidence: 99%

Analysis of Enhanced Built-up and Bare Land Index (EBBI) in the Urban Area of Yangon, Myanmar

Tin¹,

Muttitanon

2021

IJG

View full text Add to dashboard Cite

In planning and reviewing changes in the ground overview data, land distribution guidelines and identification of changes are critical. The availability of free global and historical satellite images offers a valuable resource for the built-up region to be continuously and accurately mapped and tracked year by year. For thirty years of data, this study uses Landsat images to obtain substantial and land spread data that is extremely useful for urban arrangement. This paper mainly focuses on the basic extraction of the built-up area for the urban planning area every five years from the satellite images of LANDSAT 5,7,8 and Sentinel 2A from USGS. The goal is to evaluate the year-by-year shift in the urban built-up area and to obtain the accuracy of the mapping of built-up and bare land areas in the study of the urban built-up trend from 1990 to 2020. In this research, GIS tools such as raster calculator and built-up region modeling are used to measure the indices that include the Enhanced Built-up and Bareness Index (EBBI), the Normalized Difference Built-up Index (NDBI) and the Urban Index (UI) or the Built-up Index (BUI). This study will therefore point out a variable approach to mapping traditional enhanced built-up and bare land changes (EBBI) automatically with simple indices and according to index outputs. The uncoordinated areas of land and population urbanization spread from areas and gradually the link between the expansion of urban land development and population growth has moved from weak positive to strong decoupling. The advantage of the method the enhanced built-up and bareness index (EBBI) can therefore be realized with the correlation of linear regression slightly expanded in 2020 over the last thirty years. The percentage of the outputs between the indexes and population rate was to use the entire spectral range of Landsat imageries which cause less spectral confusion between built-up area changes and higher accuracies compared to other indices. The modelling method was effective, quickly simple to implement, and can be used to find out the built-up area extraction.

show abstract

“…Village-Finder (Murtaza et al (2009)) used frequency and color features to train weak-classifiers which are then combined through Adaboost. Bramhe et al (2018) fine-tuned the VGG (Simonyan & Zisserman (2015)) and Inception-V3 (Szegedy et al (2016)) over the sentinel-2 from the satellite imagery for the extraction of built-up areas in sentinel-2 imagery. Similarly, (Tian et al (2018)) employed a strategy similar to bag-of-visual words, where the visual dictionary is built by extracting features from deep convolutional neural network (CNN).…”

Section: Related Workmentioning

confidence: 99%

Weakly Supervised Domain Adaptation for Built-up Region Segmentation in Aerial and Satellite Imagery

Iqbal¹,

Ali²

2020

Preprint

View full text Add to dashboard Cite

This paper proposes a novel domain adaptation algorithm to handle the challenges posed by the satellite and aerial imagery, and demonstrates its effectiveness on the built-up region segmentation problem. Built-up area estimation is an important component in understanding the human impact on the environment, effect of public policy and in general urban population analysis. The diverse nature of aerial and satellite imagery (capturing different geographical locations, terrains and weather conditions) and lack of labeled data covering this diversity makes machine learning algorithms difficult to generalize for such tasks, especially across multiple domains. Re-training for new domain is both computationally and labor expansive mainly due to the cost of collecting pixel level labels required for the segmentation task. Domain adaptation algorithms have been proposed to enable algorithms trained on images of one domain (source) to work on images from other dataset (target). Unsupervised domain adaptation is a popular choice since it allows the trained model to adapt without requiring any ground-truth information of the target domain. On the other hand, due to the lack of strong spatial context and structure, in comparison to the ground imagery, application of existing unsupervised domain adaptation methods results in the sub-optimal adaptation. We thoroughly study limitations of existing domain adaptation methods and propose a weakly-supervised adaptation strategy where we assume image level labels are available for the target domain. More specifically, we design a built-up area segmentation network (as encoder-decoder), with image classification head added to guide the adaptation. The devised system is able to address the problem of visual differences in multiple satellite and aerial imagery datasets, ranging from high resolution (HR) to very high resolution (VHR), by investigating the latent space as well as the structured output space.A realistic and challenging HR dataset is created by hand-tagging the 73.4 sq-km of Rwanda, capturing a variety of build-up structures over different terrain. The developed dataset is spatially rich compared to existing datasets and covers diverse built-up scenarios including built-up areas in forests and deserts, mud houses, tin and colored rooftops. Extensive experiments are performed by adapting from the single-source

show abstract

Extraction of Built-Up Areas Using Convolutional Neural Networks and Transfer Learning From Sentinel-2 Satellite Images

Cited by 18 publications

References 38 publications

Convolutional neural networks for global human settlements mapping from Sentinel-2 satellite imagery

Convolutional neural networks for global human settlements mapping from Sentinel-2 satellite imagery

Analysis of Enhanced Built-up and Bare Land Index (EBBI) in the Urban Area of Yangon, Myanmar

Weakly Supervised Domain Adaptation for Built-up Region Segmentation in Aerial and Satellite Imagery

Contact Info

Product

Resources

About