Verkada’s people analytics solution enhances our existing people detection features with face search, person attributes, person history, and person of interest (POI) events. These features give you the ability to identify and search for individuals in your camera footage, allowing you to quickly find meaningful events and save time during critical investigations.
To use Verkada’s people analytics features, no additional software or hardware is needed. This guide will show you how to optimize people analytics performance on your Verkada cameras.
To enable People Analytics on your camera follow this guide here.
The table below shows the maximum recommended distances for People Recognition and Face Recognition on our camera models.
Privacy
Verkada’s people analytics solution is built with privacy and security in mind. The people search and history results are always kept within your organization and are securely encrypted at rest and in transit.
Verkada is also committed to providing customers with full visibility into and control over their security systems. People analytics features are disabled by default and must be enabled by an administrator. Robust audit logs enable customers to track usage of and changes to their environment in order to ensure compliance with regulation and internal policies. If a more granular level of privacy is needed, cameras can also be configured with privacy regions to block certain regions of the scene from being recorded or used with people analytics.
Person of Interest Fine Tuning
Verkada’s Person of Interest function uses the distance between someone’s eyes, ears, nose, and mouth to identify unique individuals on the profile tab. The more data that gets feed to the system, the better the performance will be.
Given that Command will work off the data set that you feed it (building out profiles with images of individuals), it’s crucial to ensure your profiles are built out correctly without any pictures that are grainy, cover most facial features, or are side profiles of the person’s face.
Adding photos to a profile can be done in 1 of 3 ways. The first way is by adding photos that are generated automatically by a Verkada camera. The second way is by uploading a photo from your computer that is either in JPG or PNG format and no larger than 15 MB. The third way is through our mobile app.
Adding Photos Generated Automatically
Method 1
Select Cameras from the All Products menu
Select the People tab from the left-hand menu
Find the hyperzoom of the individual you want to add photos for
Select the three dots in the top right-hand corner of the hyperzoom
Select Create a profile
Method 2
Select a camera you want to add a profile from
Select the People tab under the camera's live feed
Select the Faces tab
Select the three dots in the top right-hand corner of the face
Select Create a profile
Adding Photos from Computer
Select Cameras from the All Products menu
Select the People tab from the left-hand menu
Select the Profiles tab
Select Create a profile in the top right-hand corner
Upload a photo to create the profile
Adding Photos via Mobile App
Select the search magnifying glass (second icon in the bottom menu)
Click the search bar at the top of the screen
Select Upload Photo
Select Camera to upload a live shot using the phone's camera or Photos to choose from your camera roll.
How to get the best POI results
Camera positioning is key, please refer to this help article as a starting point for camera mounting.
Add clear head-on shots of faces to the person’s profile. Avoid adding pictures of side profiles where only have of the face is captured.
Add face results that get generated from Verkada cameras. Generally, the photos collected by Verkada cameras themselves are great starting points for building out profiles for individuals.
Examples of Good POI profiles:
Examples of Poor POI profiles
Avoid adding pictures with people wearing:
Glasses
Masks
Hats that cover too much of the eyes
Make the best image the ‘feature photo’. The feature photo is the photo that is showcased on the main Profiles page when scrolling through the People list. The feature photo can be changed by clicking the largest photo to the left of the profile.
Avoid adding photos that are too grainy or too dark. Photos that clearly show facial features perform better than photos that are dark. People analytics generally performs best with a minimum light level of 250 lux.