Skip to content

Latest commit

 

History

History
8 lines (5 loc) · 875 Bytes

File metadata and controls

8 lines (5 loc) · 875 Bytes

Detecting Building Changes between Two Point Clouds using Patch-based CNN

This project aims at detecting building changes between two airborne point clouds obtained from two epochs. The point clouds may from airborne laser scanning or dense image matching (i.e. digital photogrammetry). The results are patch-based change masks.

The code was written in Jupyter Notebook with PyTorch framework. It contains the following architectures: FF-HH, FF-HHC, FF-DC, PSI-HH, PSI-HHC, and PSI-DC. You may adjust the hyper-parameters or network configurations for your own data. Please check the referred paper below for more details.

References

Zhang, Z., Vosselman, G., Gerke, M., Persello, C., Tuia, D. and Yang, M.Y., 2019. Detecting building changes between airborne laser scanning and photogrammetric data. Remote sensing, 11(20), p.2417. https://doi.org/10.3390/rs11202417