Inferring Contextual Data from Real-World Photography

Tiago S. Costa, Maria Teresa Andrade, Paula Viana

Abstract: The ability to collect complementary information when acquiring photographs, for instance, data and time, has gained much attention since the introduction of digital photography, especially due to the potential impact in the organization of photographic catalogues and archives. Much effort has been put to understand the types of metadata that would be useful for people when using digital images. However, scarce attention has been paid to the use of metadata to enable creative editing work with the photograph. With the definitive transition to digital in image acquisition and editing workflows, it became viable to introduce metadata throughout the complete content value chain with the potential to automate or assist the creative work. In this paper, we present eCAT, a solution developed in the framework of the FotoInMotion project that offers two contributions towards the fulfillment of such aim. Firstly, it defines tools that seamlessly acquire multiple types of contextual data from a smartphone’s built-in sensors at the time the photograph is captured, and process such data to infer physical conditions at that moment and place. Secondly, it devises methods to fuse that meta-information with tags associated to the photograph and to obtain additional knowledge from external sources that can be used in the creative process. Together, these contributions pave the way for the development of metadata-assisted creative workflows, offering context acquisitioning mechanisms to photographers to enrich their work and build engaging multimedia stories based on a still image.