Not every detected face is equal. We assess faces at the server level based on quality metrics such as pose, sharpness, brightness, pixel variance and size, and classify them as high or low quality.
After pairing a camera in Nirovision, you can get real-time feedback about the quality of the faces your camera is detecting using the calibration tool.
Nirovision displays all your recognition activity by default, regardless of its quality ⚙️. Tap on the Confidence filter in the Activity view to hide low quality events. Nirovision classifies your activity as Matched if there is a positive recognition result, or Unmatched, if we couldn't find an identity for the person in frame ⚙️. Tap on the People filter in the Activity view and choose the level of detail.
Tap on the three dots next to any recognition event in the Activity view to edit it: create a new identity from it, mark it as unknown or change it to the correct identity if it's a mistake.
If you'd like help setting up any of these new features, get in touch with us 📣
🇦🇺 Phone and email support during AEST office hours.
📚 Helpful resources from our knowledge base.
💬 Book time with us to discuss your challenges and get advise.