Abstract
Capturing geographic information from a mobile platform, a method known as mobile mapping, is today one of the best methods for rapid and safe data acquisition along roads and railroads. The digitalization of society and the use of information technology in the construction industry is increasing the need for structured geometric and semantic information about the built environment. This puts an emphasis on automatic object identification in data such as point clouds. Most point clouds are accompanied by RGB images, and a recent literature review showed that these are possibly underutilized for object identification. This article presents a method (image-based point cloud segmentations – IBPCS) where semantic segmentation of images is used to filter point clouds, which drastically reduces the number of points that have to be considered in object identification and allows simpler algorithms to be used. An example implementation where IBPCS is used to identify roadside game fences along a country road is provided, and the accuracy and efficiency of the method is compared to the performance of PointNet, which is a neural network designed for end-to-end point cloud classification and segmentation. The results show that our implementation of IBPCS outperforms PointNet for the given task. The strengths of IBPCS are the ability to filter point clouds based on visual appearance and that it efficiently can process large data sets. This makes the method a suitable candidate for object identification along rural roads and railroads, where the objects of interest are scattered over long distances.
Highlights
The use of information technology is increasing in society and with this comes an increased demand for structured information about the built environment
It is clear that the prediction for the left sample is better, but the fully convolutional network (FCN) still manages to predict a large portion of the game fence in the right sample even though the wiring is not visible
The results show that the accuracy of the classification is higher in the IBPCS subset compared to the full point cloud
Summary
The use of information technology is increasing in society and with this comes an increased demand for structured information about the built environment. The fully convolutional network (FCN) (Long et al, 2015) is an adaptation of a conventional CNN that performs semantic segmentation (pixelwise classification) instead of classifying entire images This is accomplished by replacing the last fully connected layer of the CNN, which maps the output from the last hidden layer to a vector representing the different classes, with yet another convolutional layer of size 1×1 and with a depth corresponding to the number of classes. This procedure is known as transfer learning This means that it is possible to copy the architecture of a top-performing CNN, initialize it with the weights it has learned from ImageNet, and retrain the topmost layers on a much smaller data set without over-fitting to the small data set. This makes CNNs more viable in real-world scenarios, as labeled training data typically is hard to find and time consuming to produce
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
