Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Change detection needs change information: improving deep 3D point cloud change detection

Iris de Gélis, Thomas Corpetti, Sébastien Lefèvre

TL;DR

This study focuses on change segmentation using raw 3-D point clouds directly to avoid any information loss due to the rasterization processes and proposes three new architectures to address 3-D PC change segmentation: OneConvFusion, Triplet KPConv, and Encoder Fusion SiamKPConv.

Abstract

Change detection is an important task that rapidly identifies modified areas, particularly when multi-temporal data are concerned. In landscapes with a complex geometry (e.g., urban environment), vertical information is a very useful source of knowledge that highlights changes and classifies them into different categories. In this study, we focus on change segmentation using raw three-dimensional (3D) point clouds (PCs) directly to avoid any information loss due to the rasterization processes. While deep learning has recently proven its effectiveness for this particular task by encoding the information through Siamese networks, we investigate herein the idea of also using change information in the early steps of deep networks. To do this, we first propose to provide a Siamese KPConv state-of-the-art (SoTA) network with hand-crafted features, especially a change-related one, which improves the mean of the Intersection over Union (IoU) over the classes of change by 4.70%. Considering that a major improvement is obtained due to the change-related feature, we then propose three new architectures to address 3D PC change segmentation: OneConvFusion, Triplet KPConv, and Encoder Fusion SiamKPConv. All these networks consider the change information in the early steps and outperform the SoTA methods. In particular, Encoder Fusion SiamKPConv overtakes the SoTA approaches by more than 5% of the mean of the IoU over the classes of change, emphasizing the value of having the network focus on change information for the change detection task. The code is available at https://github.com/IdeGelis/torch-points3d-SiamKPConvVariants.

Change detection needs change information: improving deep 3D point cloud change detection

TL;DR

This study focuses on change segmentation using raw 3-D point clouds directly to avoid any information loss due to the rasterization processes and proposes three new architectures to address 3-D PC change segmentation: OneConvFusion, Triplet KPConv, and Encoder Fusion SiamKPConv.

Abstract

Change detection is an important task that rapidly identifies modified areas, particularly when multi-temporal data are concerned. In landscapes with a complex geometry (e.g., urban environment), vertical information is a very useful source of knowledge that highlights changes and classifies them into different categories. In this study, we focus on change segmentation using raw three-dimensional (3D) point clouds (PCs) directly to avoid any information loss due to the rasterization processes. While deep learning has recently proven its effectiveness for this particular task by encoding the information through Siamese networks, we investigate herein the idea of also using change information in the early steps of deep networks. To do this, we first propose to provide a Siamese KPConv state-of-the-art (SoTA) network with hand-crafted features, especially a change-related one, which improves the mean of the Intersection over Union (IoU) over the classes of change by 4.70%. Considering that a major improvement is obtained due to the change-related feature, we then propose three new architectures to address 3D PC change segmentation: OneConvFusion, Triplet KPConv, and Encoder Fusion SiamKPConv. All these networks consider the change information in the early steps and outperform the SoTA methods. In particular, Encoder Fusion SiamKPConv overtakes the SoTA approaches by more than 5% of the mean of the IoU over the classes of change, emphasizing the value of having the network focus on change information for the change detection task. The code is available at https://github.com/IdeGelis/torch-points3d-SiamKPConvVariants.
Paper Structure (12 sections, 6 equations, 7 figures, 5 tables)

This paper contains 12 sections, 6 equations, 7 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Incorporating change information in deep models
  3. Considering hand-crafted features
  4. New models for 3D point cloud change detection
  5. Results
  6. Datasets
  7. Experimental settings
  8. Experimental results
  9. Results of adding hand-crafted features to the Siamese KPConv network
  10. Results of the Siamese KPConv evolution
  11. Discussion
  12. Conclusion

Figures (7)

  • Figure 1: OneConvFusion architecture for the 3D PC change segmentation. The links between successive layers were omitted for brevity.
  • Figure 2: Triplet KPConv architecture for 3D the PC change segmentation.
  • Figure 3: Encoder Fusion SiamKPConv architecture for the 3D PC change segmentation.
  • Figure 4: Influence on per class IoU of adding hand-crafted features along with 3D point coordinates as the input to the Siamese KPConv. For the 'new building', 'demolition' and 'missing vegetation' classes, the high disparity in the IoU demonstrated that adding the hand-crafted features as input had a larger influence compared to those on classes where the results were grouped around the same value.
  • Figure 5: Influence on per class IoU of the three Siamese KPConv evolutions, namely OneConvFusion, Triplet KPConv and Encoder Fusion SiamKPConv. The results of Siamese KPConv with 10 hand-crafted input features were also included for comparison purposes.
  • ...and 2 more figures