Training an AI for surface analysis

Description

This project focuses on training a Point Net model to classify geographical data from an outdoor environment which includes labelling of cars, trees, buildings, and other objects. The training utilizes a PointNet implementation to process the data, which is provided in LiDAR format. This includes around 6M cloud points (Images of the data cannot be shared publicly due to confidentiality); however, the dataset was manually labeled using MATLAB and CloudCompare to ensure precise model training and analysis.