Aplitop is investigating the application of Machine Learning for the segmentation and classification of point clouds
The capture of massive data and its representation as point clouds is revolutionizing the procedures for the elaboration of surveying works in infrastructure construction projects.
However, this huge volume of data requires additional processing – or in many cases, manual editing – to extract the information useful for engineering projects. The commercial applications currently available offer some utilities, although the results can be greatly improved.
Aplitop is researching the application of Artificial Intelligence and Machine Learning techniques for the segmentation and classification of point clouds generated for surveying projects.
Segmentation or clustering is an unsupervised learning technique that consists of dividing the points that have common characteristics into groups, without assigning semantics to each one. This would facilitate the selection of objects that should be ignored for the generation of the digital terrain model.
Classification is a supervised learning technique consisting of assigning a category to each point from a previously trained model, in order to subsequently perform the automatic extraction of entities.