Object detection in airborne laser scanning (ALS) data using deep learning

In partnership with the Lower Saxony State Office for Preservation of Historic Monuments, we are developing a method for automatically detecting archaeological objects in airborne laser scanning data. The type of objects to be detected are mainly those of interest by archaeologists such as heaps, shafts, charcoal piles, pits, barrows, bomb craters, hollow ways, etc. They could be point, linear, or areal objects. To this end, we are using deep learning techniques; namely, convolutional neural networks (CNNs) to classify height images from the region of interest. A combination of multiple (in most cases 5) CNN classifiers are then used to detect and localize objects of interest in a digital terrain model acquired from the region of interest.

Tasks: 

1. Data pre-processing: The first task in this project is pre-processing raw ALS data to generate digital terrain models (DTM) and create training/test data for our convolutional neural networks.
2. Classification: The second task is to train CNN models to take squared height images (n x n) as inputs and giving a class label (among the objects of interest) as an output. The value of n is to be searched and optimized for best results.
3. Detection: The next task in this project is to use the best (5) classifiers in the previous step to scan through a big area of interest clipping small height images (of size n x n), classify them and finally generate a heat map for each object class showing the location of each of them in the region.
4. Change detection: In order to preserve historic monuments, it is important to notice the changes they undergo through years, investigate and try to protect them from going bad. This step is allocated for studying and detecting the changes in the region of interest.
5. Complete software package and deployment: The ultimate goal of this project is to integrate all the previous step in one single software package and deploying it for automatic usage.