FAQs

Nirovision is an identity recognition platform that accurately identifies and tracks individuals and their attributes from security camera footage. It’s designed to provide businesses with insights to help improve workplace health and safety, aid decision making and optimise operations.

Nirovision leverages both an on-premise server and the cloud, along with sleek desktop and mobile applications to view and derive insights from camera footage.

This is what is required:

• Broadband / Cable / Fiber internet connection
• IP camera(s) strategically placed for facial recognition
• A dedicated computer server running the Nirovision software
• The Nirovision phone app for iOS or Andriod

Nirovision’s algorithm identifies facial features from security camera footage by extracting embeddings, or key features, from an image of the subject’s face. This can include things like the relative position, size, and/or shape of the eyes, nose, cheekbones, and jaw. These features are then compared to a database of stored images to find matches. To confirm a match, Nirovision uses quality metrics to provide a verdict – a verdict that relies on thresholds which are configurable based on your particular circumstances – to improve the accuracy.

Nirovision scores 99.6% accuracy on the the Labeled Faces in the Wild dataset, a public benchmark for face verification. However accuracy is determined by so many factors such as the placement of cameras, quality of images, and lighting. For this reason, we don’t like giving accuracy numbers, as a very weak model can get 100% accuracy on an easy dataset. The more quality images added to a dataset, the better Nirovision’s ability to match a face with an identity. We work with our customers to improve and strengthen results over time.

It really depends on the location of the implementation, as different laws have different levels of information requirements.

In Australia, there is no specific legislation regulating the use of facial recognition technology per-se, but there are federal and state laws that regulate the use of CCTV in private property, workplaces, and liquor-selling venues, and the handling of Australian’s personal information.

Your on-premise server stores video footage as a byproduct of processing your IP cameras stream through our facial recognition pipeline.

Video is stored on the device for later access by the cloud infrastructure or end user applications, as required.

Anytime you play video from the mobile and webapps, that video is streamed from the server (directly connected to it if accessed on the same network, and tunnelled through a VPN if accessed remotely).

The server has rolling-delete functionality, which automatically deletes old footage once you reach 80% HDD capacity.

All our cloud interactions are protected by OAuth 2.0 authentication and HTTPS protocols.

Our authentication partner is Auth0, an industry leading identity management platform.

Our cloud infrastructure provider is AWS, with all data being stored and transmitted solely in the AWS Sydney Availability Zone. In 2019, the Australian Cyber Security Centre (ACSC) certified AWS for hosting Australian Government data classified up to the PROTECTED classification level. 

Nirovision integrates with selected thermal cameras which on their own are able to detect body temperature. However Nirovision takes this a step further by matching an identity and/or face to a specific reading, and comparing an individual’s current temperature to their average normal baseline temperature over time. If the individual is above their baseline a notification can be sent to selected Nirovision users such as HR or security.

Baseline temperatures are precise to the hour, as it is common for the human body to get warmer throughout the day. If an hourly baseline is not available, the individual’s median is used instead.

Contact us to see a free demonstration.