In this section:
Consider enabling Deduplication to exclude coinciding facial recognition events within one camera group.
To enable event deduplication, do the following:
- Enable the offline mode of video face detection for each camera in the group. See Add Camera for details.
- Navigate to the Preferences tab. Click Camera Groups.
- Open the camera group settings.
- Check Deduplicate Events and specify the deduplication interval in seconds.
The deduplication algorithm works as follows:
- In the offline mode, the server receives one best face snapshot per tracking session on a camera (a tracking session continues until a face disappears from the camera field of view).
- If within the same camera group, there are several tracking sessions on a camera(s) within the specified deduplication interval, FindFace Security will handle the received snapshots in the following way:
- If there is a match with a dossier within the preceding deduplication interval, FindFace Security drops a newly acquired snapshot. Otherwise, the snapshot is saved to the database.
- For unmatched faces, when performing deduplication, FindFace Security considers both similarity between faces in the snapshots and face quality. FindFace Security drops all snapshots of similar faces within the deduplication interval unless a new face is of higher quality. This guarantees that the system deduplicates events without skipping high-quality faces, which are essential for further video analytics.