Best Practices for AI – Rhombus Support

Topics included in this article:

Overview
Initial AI Expectations
Camera Setup for Optimal AI Accuracy
- Camera Positioning
Placing Activity Regions
Human Movement
Facial Recognition
Vehicle Movement
License Plate Recognition
Helpful Links
Contact Support or Sales

Overview

Rhombus cameras use artificial intelligence to enhance security monitoring by automatically detecting and tracking key objects in view. The AI can identify humans, faces, vehicles, and license plates in real time. All processing is securely handled through Rhombus' cloud infrastructure, enabling tracking, streamlined searches, and actionable insights for users through the Rhombus Console.

Different event types must be added to the alert policy assigned to the camera in order to populate events.

Initial AI Expectations

Due to a variety of environments and applications for Rhombus cameras, the AI thresholds may need to be adjusted to fine-tune event capture. For more information, please review the AI Confidence Settings article.

Camera Setup for Optimal AI Accuracy

Specific features will have more in-depth setup instructions. For more information, please review the article for the type of AI feature you are working with.

Camera Positioning

Lighting, camera angles, distances, and camera positioning all impact the success of capturing AI events.

When mounting a camera, try to eliminate all unused space from the camera view.
Ensure all cameras are mounted at the appropriate distance for the camera model and AI detection type you are trying to capture. Follow the Optics and Object Distances for Analytics guide for additional information.
Ensure there is sufficient lighting while also avoiding backlighting for features like facial recognition and license plate recognition.

Placing Activity Regions

For camera views that have high traffic, like in a public space, using activity regions can help reduce the number of events that are created.

Activity regions can be used in two ways:

Set an area of interest where you want to capture events. Ex.) The entrance to a building.
Set up regions where you want to ignore event capture. Ex.) An area that is predominantly foliage or a public road.

To learn more, visit the Configuring Activity Regions article.

Human Movement

Human movement detection requires a camera position where most of the human body moves through the frames for AI identification.

Figure 1.) Human movement is captured on a device as someone walks by. This event will also populate on the timeline.

Facial Recognition

Facial recognition occurs after human movement has been identified by the system. If the camera is not setup for optimal human movement capture, it will not be able to progress to the facial recognition step. The camera has additional placement requirements that assist with optimizing facial recognition.

For more information, please review the Managing Facial Recognition article.

Vehicle Movement

It is recommended to have the camera positioned so the full car enters the field of view. If only a portion of the car enters the field of view, vehicle movement detection may not trigger. How fast a vehicle moves through the frames also impacts whether the AI can process. Ensuring there is sufficient time of the vehicle moving through the field of view will help optimize the accuracy of detections.

License Plate Recognition

License plate recognition occurs after vehicle movement has been identified by the system. The camera has additional placement requirements that assist with optimizing license plate capture.

For more information, please review the Managing License Plate Recognition (LPR) article.

Helpful Links

Contact Support or Sales

Have more questions? Contact Rhombus Support at +1 (877) 746-6797 option 2 or support@rhombus.com.

Interested in learning more? Contact Rhombus Sales at +1 (877) 746-6797 option 1 or sales@rhombus.com.