Our group designs algorithms and software architectures for visual inference: converting raw image data into semantic understanding. Our inspiration is biology; our “target” applications range from assistive devices [Rivera-Rubio et al, 2013] to medical images [Cao et al, 2011] through to images that are contained within museums or so-called heritage collections [Creswell et al, 2016]. Recently, we have started to work towards robotic applications, particularly where the actions of a robot are informed by visual information [Arulkumaran et al, 2016].