This provides the ability to compare an image against a database of images and display matches above a user-defined threshold, even if the images are different. This can be the case due to cropping, resaving, reformatting, etc, or due to different images having similar objects within them.
In the above example f2 can find a similar, but not necessarily identical, piece of clothing in a crowd. Matches can be determined on the basis of shape, or a mix of color and shape. This means that crowds can be scanned and individuals of potential interest highlighted for officers to follow up. When combined with f2’s advanced face recognition technology this provides a powerful means of identifying people who are on a watchlist. If required the process can be significantly automated – a camera focused on a busy sidewalk can display images of people who match someone on a watchlist and highlighting this for an officer at a control centre to verify and, if positive, inform other officers close to the scene to take action.