Locate a subject without face recognition

Find specific subjects in security footage without the need of using biometric data, which may not be allowed by certain legislations.

With privacy regulations like GDPR and others in place and tightening the current legal landscape regarding the transmission of personally identifiable information, biometric readings of people (i.e. faces, prints, voice, etc.) it is ever more pressing to facilitate Law Enforcement’s work providing alternative identification methods that yield good results and help in their investigations.

Intelion provides identifications by similarity, for example referred to clothing. This way it is possible to identify a person among many others just by the clothes they are wearing. Even with this method not being perfect, it helps narrowing down suspects or tracking them amongst security or traffic cameras.

Search by clothing similarity

One of the methods to be able to track someone in a crowd or between security cameras, is by the clothing he or she is wearing. This does not require biometric readings of any kind and thus does not contravene regulations like the General Data Protection Regulation. On the other hand, it allows Law Enforcement a quite reliable tracking of a suspect, at least until a court order allows the face recognition to be applied.

In Intelion it is possible to filter people by the top and bottom color they are wearing, which allows some form of tracking, at least distinguishing and finding suspects in a crowd. Obviously, several people with the same colors can trigger the filter, but here is where the law enforcement officer experience comes in, selecting the right person (by hair color, by gate, style of clothing, etc.).

The filters in Intelion allow a granular search of different people or object features.

Intelion identification filters

The identification thumbnails help see the results in a very graphical and easy way.

The different analytics modules than can be integrated into the Intelion platform yield different results on the incoming media files (in this case video from security or traffic cameras). As all this information is in the form of metadata, and the analyzers have different criteria depending on the vendor, they yield different results with a wide range of granularity. Fortunately, Intelion combines all the data into a comprehensible interface, that allows applying several filters to the incoming results.

For example, the analyzers able to detect people usually can differentiate between the color they are wearing on the lower part, and the color on the upper part. This is one way of tracking someone without resorting to more detailed biometrics analysis. Simply selecting the desired color in both listboxes automatically shows the matching results on screen, in the form of thumbnails, so it is easy to dismiss the persons that match the colors but are not the suspect. A click on any thumbnails brings up that specific point in the video to the see the context and corroborate the identification.

Geographic pintpointing

Once the positive identification has been made, Intelion provides a map view to help locate the identification in physical space. As most security imaging devices either have an integrated GPS or, if they are fixed in place, have their coordinates programmed in, Intelion can recover this information and show the identifications on a map. Together with the identification timestamp this is a clear way to identify and find someone, even with no biometric data travelling from a camera to the system.

In cases where immediate action is required and the officers cannot wait for a court order to be issued, this additional form of identification can help locate someone and get him/her into custody, pending further confirmations.

The map shows where the identifications have taken place, which can be combined with the timestamp.

Watch Intelion in action

See how Intelion can find people by clothing color only.

In this short video Intelion shows how to use Intelion’s Artificial Intelligence Video Analytics to find people by their clothing in direct surveillance camera or VMS feeds, without additional information, in order to not use facial biometrics due to legal restrictions in some areas.

Conclusion

When the use of biometric identifications (for example via facial recognition) are out of the question due to legal and privacy law reasons, there are still ways to find and follow a suspect with the help of security cameras, and at the same time adhering to all the applicable rules and laws. The filtering by clothing color, split in top and bottom halves, is one ways that result quite accurate, speeding a time sensible case forwards, until a court order arrives that may grant biometric identification techniques.