Building Generalization Using Deep Learning

Sester, Monika; Feng, Yilin; Thiemann, Frank

doi:10.5194/isprs-archives-xlii-4-565-2018

Cited by 32 publications

(15 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results from our previous study [29] show that the parallelism and rectangularity of the building patterns in some cases can not be preserved adequately. Another neural network architecture, which has great potential to tackle this problem, is the GAN [33].…”

Section: Generative Adversarial Network (Gan)mentioning

confidence: 78%

“…In this paper, three network architectures are applied for building generalization. This is a substantial extension of a previous work [29], where only one architecture was employed. The idea of that paper was inspired by the work from Simo-Serra [30], where sketch drawings were simplified using a deep convolutional neural network.…”

Section: Approachmentioning

confidence: 84%

“…The models are able to generate less blurring results [35]. This is also the basic network architecture used in our previous work [29]. In that work, two modifications have been made compared with the original U-net [31].…”

Section: U-netmentioning

confidence: 99%

See 2 more Smart Citations

Learning Cartographic Building Generalization with Deep Convolutional Neural Networks

Feng

Thiemann

Sester

2019

IJGI

View full text Add to dashboard Cite

Cartographic generalization is a problem, which poses interesting challenges to automation. Whereas plenty of algorithms have been developed for the different sub-problems of generalization (e.g., simplification, displacement, aggregation), there are still cases, which are not generalized adequately or in a satisfactory way. The main problem is the interplay between different operators. In those cases the human operator is the benchmark, who is able to design an aesthetic and correct representation of the physical reality. Deep learning methods have shown tremendous success for interpretation problems for which algorithmic methods have deficits. A prominent example is the classification and interpretation of images, where deep learning approaches outperform traditional computer vision methods. In both domains-computer vision and cartography-humans are able to produce good solutions. A prerequisite for the application of deep learning is the availability of many representative training examples for the situation to be learned. As this is given in cartography (there are many existing map series), the idea in this paper is to employ deep convolutional neural networks (DCNNs) for cartographic generalizations tasks, especially for the task of building generalization. Three network architectures, namely U-net, residual U-net and generative adversarial network (GAN), are evaluated both quantitatively and qualitatively in this paper. They are compared based on their performance on this task at target map scales 1:10,000, 1:15,000 and 1:25,000, respectively. The results indicate that deep learning models can successfully learn cartographic generalization operations in one single model in an implicit way. The residual U-net outperforms the others and achieved the best generalization performance.

show abstract

Section: Generative Adversarial Network (Gan)mentioning

confidence: 78%

Section: Approachmentioning

confidence: 84%

See 1 more Smart Citation

Learning Cartographic Building Generalization with Deep Convolutional Neural Networks

Feng

Thiemann

Sester

2019

IJGI

View full text Add to dashboard Cite

show abstract

“…Over the decades, several approaches have been proposed for modeling different stages of building generalization. Some of them focus on modeling the overall process of building generalization using rule‐based (Harrie & Weibel, 2007), agent‐based (Renard, Gaffuri, Duchêne, & Touya, 2011), optimization (Sester, 2005), or machine learning (Sester, Feng, & Thiemann, 2018) approaches. Some others include pre‐processing steps in building generalization.…”

Section: Literature Reviewmentioning

confidence: 99%

A new model combining building block displacement and building block area reduction for resolving spatial conflicts

Pilehforooshha

Karimi

Mansourian

2021

Transactions in GIS

View full text Add to dashboard Cite

Applying pre‐processing and geometric transformation, for the generalization of building blocks, can lead to spatial conflicts which are mainly resolved by displacement. However, the conflicts may not be resolved, due to the insufficient space for displacement or the existence of other objects that prevent displacement. This article proposes a novel model combining building block displacement and building block area reduction for resolving spatial conflicts. In this model, first the immune genetic algorithm with improved objective function, with the goal of minimizing the total conflicting area, is used for displacement. Second, a building block area reduction model is applied to reduce the area of the unresolved conflicting building blocks. For this, a building segment partitioning model is developed for partitioning the boundaries of conflicting building blocks. Then, the conflicting segments are locally simplified for more adaptation to the initial generalized ones. We generalized two datasets from scale 1:25 to 1:50k and then used the datasets with scale 1:50k to assess the proposed model. The results demonstrate that the proposed model has improved the correctness and completeness by 3.62 and 4.24% for dataset 1 and 5.67 and 7.92% for dataset 2, respectively, compared to one of the recent models for resolving conflicts.

show abstract

“…Deep learning techniques have shown state-of-the-art performance in automatic target recognition, surface classification, change detection, and feature extraction [31][32][33]. A major problem is related to the collection of necessary labeled data.…”

Section: Introductionmentioning

confidence: 99%

Detection of Collapsed Bridges from Multi-Temporal SAR Intensity Images by Machine Learning Techniques

2021

View full text Add to dashboard Cite

Bridges are an important part of road networks in an emergency period, as well as in ordinary times. Bridge collapses have occurred as a result of many recent disasters. Synthetic aperture radar (SAR), which can acquire images under any weather or sunlight conditions, has been shown to be effective in assessing the damage situation of structures in the emergency response phase. We investigate the backscattering characteristics of washed-away or collapsed bridges from the multi-temporal high-resolution SAR intensity imagery introduced in our previous studies. In this study, we address the challenge of building a model to identify collapsed bridges using five change features obtained from multi-temporal SAR intensity images. Forty-four bridges affected by the 2011 Tohoku-oki earthquake, in Japan, and forty-four bridges affected by the 2020 July floods, also in Japan, including a total of 21 collapsed bridges, were divided into training, test, and validation sets. Twelve models were trained, using different numbers of features as input in random forest and logistic regression methods. Comparing the accuracies of the validation sets, the random forest model trained with the two mixed events using all the features showed the highest capability to extract collapsed bridges. After improvement by introducing an oversampling technique, the F-score for collapsed bridges was 0.87 and the kappa coefficient was 0.82, showing highly accurate agreement.

show abstract

Building Generalization Using Deep Learning

Cited by 32 publications

References 18 publications

Learning Cartographic Building Generalization with Deep Convolutional Neural Networks

Learning Cartographic Building Generalization with Deep Convolutional Neural Networks

A new model combining building block displacement and building block area reduction for resolving spatial conflicts

Detection of Collapsed Bridges from Multi-Temporal SAR Intensity Images by Machine Learning Techniques

Contact Info

Product

Resources

About