Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Classification of Building Information Model (BIM) Structures with Deep Learning

Francesco Lomio, Ricardo Farinha, Mauri Laasonen, Heikki Huttunen

TL;DR

This work uses classical and modern machine learning methods to categorize images of building designs into three classes: Apartment building, Industrial building or Other, and compared four different methods based on classical and deep learning.

Abstract

In this work we study an application of machine learning to the construction industry and we use classical and modern machine learning methods to categorize images of building designs into three classes: Apartment building, Industrial building or Other. No real images are used, but only images extracted from Building Information Model (BIM) software, as these are used by the construction industry to store building designs. For this task, we compared four different methods: the first is based on classical machine learning, where Histogram of Oriented Gradients (HOG) was used for feature extraction and a Support Vector Machine (SVM) for classification; the other three methods are based on deep learning, covering common pre-trained networks as well as ones designed from scratch. To validate the accuracy of the models, a database of 240 images was used. The accuracy achieved is 57% for the HOG + SVM model, and above 89% for the neural networks.

Classification of Building Information Model (BIM) Structures with Deep Learning

TL;DR

This work uses classical and modern machine learning methods to categorize images of building designs into three classes: Apartment building, Industrial building or Other, and compared four different methods based on classical and deep learning.

Abstract

In this work we study an application of machine learning to the construction industry and we use classical and modern machine learning methods to categorize images of building designs into three classes: Apartment building, Industrial building or Other. No real images are used, but only images extracted from Building Information Model (BIM) software, as these are used by the construction industry to store building designs. For this task, we compared four different methods: the first is based on classical machine learning, where Histogram of Oriented Gradients (HOG) was used for feature extraction and a Support Vector Machine (SVM) for classification; the other three methods are based on deep learning, covering common pre-trained networks as well as ones designed from scratch. To validate the accuracy of the models, a database of 240 images was used. The accuracy achieved is 57% for the HOG + SVM model, and above 89% for the neural networks.

Paper Structure

This paper contains 10 sections, 7 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Data
  4. HOG + SVM
  5. Pre-trained Neural Network
  6. MobileNet
  7. ResNet
  8. Neural Network with Random Structure
  9. Results
  10. Conclusion

Figures (7)

  • Figure 1: Example of a BIM virtual representation of an apartment building as presented to the models. The image has a dimension of 224x224 pixels.
  • Figure 2: Example of the different images for each BIM structure: each image is generate from a completely different angle of the structure.
  • Figure 3: Example of augmented image as shown to the network. From left to right it can be seen: the original image, a rotation, horizontal shift, vertical shift and a horizontal flip.
  • Figure 4: Example images for each class. From left to right: an apartment building, an industrial building and a structure belonging to the class "Other".
  • Figure 5: Example of histograms of oriented gradients computed for one of the images in the dataset.
  • ...and 2 more figures