NTT DATA presents Global Open Innovation Contest 7.0


Human action recognition in the
visual surveillance domain

Banking surveillance systems should be able to recognize a potential intruder among current public in a branch or corporative buildings.
Nowadays current systems offer the detection of individuals enrolled in a blacklist or a whitelist.

The challenge is to recognize potential threats due to their
behaviour, commonly their faces never provide a clear shot because they are partially hidden, they spent a large amount of time in the premises without a clear objective, they approach people on the premises searching for an opportunity to cheat or may present a more important threat, among others.
Potentially the identified individuals should be enrolled in a list of suspects.