Recent advances in high resolution geographic data collection platforms such as Unmanned Aerial Vehicles (UAVs) have rapidly changed the way we visualize spatial information. Integrating different modalities collected by UAV can be used to indicate the state of facility operations by providing environmental context about the surroundings. However, performing image segmentation combining features from multiple sensors situated on a UAV is challenging owing to the huge volume of data, lack of co-orientation among sensors and variability in their spatial and temporal resolution. In this study, we provide a novel framework to perform image segmentation on fused LiDAR and aerial imagery data from unlabeled images using a Bayesian non-parametric clustering approach.