Unmanned Aerial Vehicle (UAV) images are increasingly being used for building footprint extraction due to their high spatial resolution, cost-effectiveness, flexibility, and accessibility. However, extracting building footprints from UAV images in unplanned dense regions of cities of developing nations such as India, is challenging. This study proposes Deep Learning (DL) approach to building footprint extraction by integrates UAV images with elevation data. The proposed approach involves development of five state-of-the art Deep Learning (DL) models on manually annotated RGBE (RGB + Elevation) imagery. We demonstrate the hypothesis that usage of elevation can be very effective in overcoming misclassifications primarily due to similar spectral information. The approach was evaluated on a dataset of UAV images acquired from a region of Bhopal city, India. The results show exceptional performance of all the models in segmenting the building footprints. We also found that the use of elevation was very significant in training and helped boost the model performances by ~6%. The proposed approach has potential applications in a variety of real-world scenarios, such as urban planning, disaster management, and environmental monitoring.