InReality’s Camera Analytics platform makes insights at scale possible. Our platform provides capabilities to gather Traffic, Impressions, Engagement, and Dwell data, and present the key data you need to take action with agility.
Camera installation is critical when gathering data. The placement, angle, camera type, and lighting are important for all types of applications.
Applications
Traffic
Camera installation for a Traffic application collects Traffic data to help you understand the traffic volume and behavior in an area (AOI) of your venue.
Traffic attributes include: Traffic Count, Dwell, Opportunity to See, Direction and Path.
Impressions
Camera installation for an Awareness subscription is designed to collect demographics data to help you determine the viewer profile, behavior, mood and volume for a specific area of interest in your venue.
Visitor attributes include: Face Count, Dwell, Gender, Age, Age Group and Mood.
Engagement
Camera installation for an Engagement subscription is designed to capture interaction data with products and screens, and to help you understand these interactions and the behavior of visitors in an area of your venue.
Engagement attributes include: Pick-Up, Reach-In, Touch, POS and 3rd Party Data.
Person Detection and Human Pose Estimation
Camera installation for Person Detection and Human Pose Estimation is designed to collect traffic data, demographics data, and interaction data with products and screens, and to help you understand these interactions and the behavior of visitors in an area (AOI) of your venue.
Attributes include: Traffic Count, Dwell, Opportunity to See, Pick-Up, Reach-In, and Touch.
Recommended Placement and Location
Many factors will determine the best placement for sensors, depending on your application.
Applications can include:
- Measuring the number of people who interact with digital signage.
- Measuring traffic in an area of your venue.
- Measuring the number of people who interact with products.
- Measuring dwell and/or traffic of people who interact with products and digital signage.
- Capturing interactions, while leveraging a variety of IoT sensors.
Placement is often dependent on the application and the structure of the area you are monitoring. For example the placement of walls, windows, lighting and displays will be a factor into sensor placement. The type of camera in use will also determine the location and placement. In sections 5 & 6 below, we will provide data to help you determine the height of camera placement, and the range of distance from the camera for measurement.
Camera Types
Standard 1080P USB cameras are most commonly used for a Traffic application. They typically provide the best performance per controller dollar and are often simpler to install. Using a camera with a wider angle, provides greater coverage. They require power and internet connection and are usually limited to 5-15 feet distance from the controller, based upon cable length, unless using a USB cable extender.
IP cameras in ceiling or wall mount configuration are also available. IP cameras require power and ethernet connection and must be placed in the correct location for appropriate line of sight. InReality customers are using multiple types of IP cameras from several suppliers.
InReality can provide a variety of IoT sensors for Engagement and Dwell installations, including NFC readers, optical sensors, and motion sensors. We also provide two different types of deep learning (DL) models for Engagement measurement, including Person Detection and Human Pose Estimation.
Camera Mounting Options
5.1- Traffic Mounting Options
Mounting the camera between 6’ - 8’ at a 5 to 20 degree angle is recommended for Traffic applications. Shown below-left is a photo taken during InReality installations. Note the camera is above eye level, and there is also has a ceiling mounted sensor for traffic measurement.
Shown above-right is a photo utilizing the ‘Person Detection’ model, which is used for Traffic applications only. In regards to CPU and memory usage, this option requires less system resources than ‘Human Pose Estimation’. It functions best at detecting people who are standing upright. For classifying an object as a person, this model needs to view 60 - 70% of the human body.
Ceiling mount cameras are commonly used for Traffic applications. Shown below are close-ups of two ceiling mount cameras, and a retail installation with a ceiling mount camera used for traffic count.
5.2- Impressions Mounting Options
Commonly cameras are mounted in areas of interest which could include: Point of Sale (POS), Digital Signage, end-caps, sales displays, and entrances/exits. They allow you to collect demographics data to help you determine the viewer profile, behavior, mood and volume for a specific area of interest.
Digital Signage screens may have a built-in camera, which is normally straight on, at eye-level with no angle. This works best for Impression and Engagement applications. Shown below are two photos taken during InReality installations. Note the cameras are approximately at eye level. On the left is a POS counter, with a camera housed in a concealed custom housing. On the right is a digital sign, with a camera, and a shelf containing products. Both cameras are circled in red, and otherwise are unseen.
5.3- Engagement and Dwell Mounting Options
With Engagement and Dwell applications, you will collect interaction data to help you understand the interactions and behavior of visitors in your venue. These areas of interest could include: digital signage, product shelves, and sales displays.
For Reach-In applications, a camera mounted above the shelf may be best. Shown below are two photos taken during InReality installations, utilizing multiple Engagement IoT sensors.
Using Person Detection and Human Pose Estimation technology, you may collect Traffic, Impression and Engagement data. Shown below are two photos taken during InReality installations, utilizing ‘Reach-In’ zones, and ‘Human Pose Estimation’ which is a deep learning technology. These two features can be used independently. In both photos, the user-defined zone is activated when the person’s hand enters the zone, updating a value field.
Recommendations - Height, Distance & Angle
Traffic
Camera installation for a Traffic application: We recommend a wall mounting height between 8 and 10 feet for Traffic count. A camera angle tilt of 10-20 degrees is recommended, which allows the camera to see the full floor area, where measurement is required. An overhead camera is also an option.
- Under normal conditions, traffic count is accurate if a person is less than 50 feet from a camera.
Impressions
Camera installation for an Impression application: We recommend a wall mounting height of between 5 and 6 feet, for Impressions & Facial Detection. An angle of tilt of 0-10 degrees is recommended. The best recommendation is a height of 62 inches with a 0 degree angle.
- Under normal conditions, impressions are accurate if a person is less than 12 feet from camera.
Engagement
Camera installation for an Engagement application: We recommend a wall mounting height of between 5 and 6 feet, for Engagement - Facial Detection. An angle of tilt of 0-10 degrees is recommended. The best recommendation is a height of 62 inches with a 0 degree angle.
- Under normal conditions, engagement is accurate when a person is up to 12 feet from a camera.
Person Detection and Human Pose Estimation
Camera installation for Person Detection and Human Pose Estimation can vary significantly. We suggest a wall mounting height between 5 and 10 feet. A camera angle tilt of 0-20 degrees is recommended.
- Under normal conditions, detection is accurate when a person is up to 50 feet from a camera.
Lighting Considerations
While lighting is very important for facial detection, it is less critical for traffic count. In facial detection applications, the camera is identifying facial landmarks such as eyes, nose and mouth, which requires good lighting and full coverage.
InReality cameras will work under normal retail lighting conditions, which is 50 footcandles or 50 lumens per square foot. Having enough light is important. Glare from sunlight through windows or from bright spotlights can also cause issues with facial detection.
When installing and mounting sensors, ensure the area used for facial recognition will have a consistent 50 lumens per square foot throughout your entire day. Test the area prior to making final installations, checking for shadows and ensuring the lighting does not change significantly during your open hours. Test your system and make lighting adjustments or adjust camera placement to ensure success.
Testing the Camera’s Field of View
During the camera installation, if you do not have an attached display screen to check the camera view and camera angle, the InReality Device Manager can be used.
8.1- Access the Device Manager
A cloud account will be provided for each user. You will be provided a User ID and Password. The same User ID and Password will be used for all InReality applications.
Access the Device Manager at: https://app.inreality.com/v3/admin/devices
Enter the provided User ID and Password.
8.2- Select the Device
The list of your devices will be displayed. Ensure the Status is ‘Online’, showing it is connected. Click on the ‘Actions’ button (), and select ‘Screenshot’.
8.3- Select the Camera
When you click the ‘Screenshot’ option, the following pop-up screen with menu is displayed. From the drop-down menu, ensure Demographics Camera is selected. Then click Refresh button (
8.4- The Camera View
After several seconds, the current camera view will be displayed.
8.5- Additional Testing
To take additional shots for testing the angle and placement of the camera, click the REFRESH button at the bottom of the image. Press the CLOSE button to close the image.