Optimizing cognitive analysis sensitivity of photospheres using cube maps

Image cognition is a fast developing field of modern technology. Considerably accurate, well trained algorithms have already been developed and exposed as public APIs for automated analysis of conventional photographs. However, in recent past, the photography field has shifted from conventional photography, embracing new concepts such as 360 surround photography or photospheres. With the advent of latest consumer level technologies such as Virtual Reality, the importance of photospheres has grown even wider as it could provide an immersive, real-life like experience to the viewer. However, the well trained image cognition implementations available today do not perform well on 360 surround photos due to the distortions in them caused by high field of view. This paper presents a unique, innovative approach, inspired from well-known concept, cube maps, to significantly improve compatibility of image cognition systems with 360 surround photography. We have obtained 2–3 fold increase in sensitivity of conventional image cognition algorithms on analyzing photospheres. Through the proposed framework, existing technologies such as Image Cognitive Analysis, Automated Surveillance and Content based indexing technologies can be made available for 360-imagery at an affordable cost.