Sim2Air - Synthetic aerial dataset for UAV monitoring

In this paper we propose a novel approach to generate a synthetic aerial dataset for application in UAV monitoring. We propose to accentuate shape-based object representation by applying texture randomization. A diverse dataset with photorealism in all parameters such as shape, pose, lighting, scale, viewpoint, etc. except for atypical textures is created in a 3D modelling software Blender. Our approach specifically targets two conditions in aerial images where texture of objects is difficult to detect, namely illumination changes and objects occupying only a small portion of the image. Experimental evaluation confirmed our approach by increasing the mAP value by 17 and 3.7 points on two test datasets of real images. In analysing domain similarity, we conclude that the more the generalisation capability is put to the test, the more obvious are the advantages of the shape-based representation.