⭐ Configuring Analytical Rules with Intelligent Mask-Based Detection

    Overview

    The analytical rule configuration feature in 3dEYE's cloud-based VMS platform provides sophisticated AI-powered object detection capabilities with unprecedented flexibility. By leveraging analytical masks, you can create highly targeted detection zones that trigger intelligent alerts only when specific conditions are met, significantly reducing false positives and ensuring that your security system responds to genuine events of interest.

    Key Benefits

    • Precision Detection: Define exact areas within your camera's field of view where object detection should occur, eliminating unwanted alerts from irrelevant activity in other parts of the frame.
    • Flexible Object Recognition: Choose from multiple object types, including person, cars, smoke, fire, PPE, LPR and loitering detection, allowing you to tailor monitoring to your specific security requirements.
    • Intelligent Loitering Detection: Configure sophisticated loitering parameters that trigger alerts only when objects remain in designated areas for specified durations with customizable occupancy thresholds.
    • Optimized Alert Management: Set confidence thresholds to control alert sensitivity, ensuring that webhooks and notifications are sent only when the AI detection meets your required certainty (confidence) level.
    • Scalable Cloud Architecture: All analytical processing occurs in the cloud, eliminating the need for specialized on-premises hardware while providing consistent performance across your entire camera deployment.

    Configuration Steps

    Step 1: Access Analytical Rule Editor

    1. Navigate to the camera you wish to configure in your 3dEYE platform
    2. Select the Analytics tab from the top navigation menu
    3. Click to create a new analytical rule or edit an existing one


    Step 2: Define Detection Masks

    Masks allow you to specify precise areas within the camera frame where detection should occur:

    1. Drag your mouse over the camera thumbnail to begin configuring your first detection zone
    2. Use your cursor to draw a polygon on the camera preview image, defining the boundary of your detection area
    3. The mask will appear as a highlighted overlay on the video feed (shown in cyan in the example)
    4. Multiple masks can be created by clicking + Add Another Mask for complex detection scenarios
    Masks can have any shape, you can add more corners simply by clicking on the center dot of any of the sides. If you made a mistake simply drag the corner to the desire position or delete the mask altogether and start over.

                   

    Step 3: Select Object Types for Detection

    In the Edit Analytical Rule interface, begin by selecting which object types the AI should detect:

    1. Click the object type dropdown menu 
    2. Choose from available detection types: Person, Car, Smoke, Fire, etc
    3. Multiple object types can be selected for comprehensive coverage
    4. Toggle can be used to enable or disable the detection for this rule
        

    Step 4: Configure Selected Targets

    Under Selected targets, specify which objects should trigger detection within your masks:

    1. Choose from available object categories: Person, Car, Smoke, Fire
    2. Select Person Loitering for advanced behavioral detection
    3. Click the X next to any object type to remove it from the detection criteria
    4. The system displays the total number of options selected for easy reference 

    Step 5: Set Loitering Parameters (If Applicable)

    For loitering detection, configure two critical parameters:

    • Loitering Duration: Enter the number of seconds an object must remain within the mask before triggering an alert. This prevents false alarms from brief, legitimate activity.
    • Loitering Objects: Specify the minimum number of detected persons that must be present simultaneously within the mask to trigger a loitering event (default: 1). Increase this number to detect group loitering or crowding scenarios.

    Step 6: Configure Trigger Event

    Select which event from the camera should trigger cloud detection:

    1. Open the Configure Trigger dropdown menu
    2. Select desired trigger type (camera must be configured to detect this event type and be able to send it over ONVIF channels)
    3. This setting determines when the cloud AI will start detecting the specified objects within your masks, it will initiate cloud AI detection processing

    ⚠️IMPORTANT: Ensure the camera is configured to detect in all areas of interest with sufficiently high sensitivity to avoid missing real events. It's better to have a few false motion or intrusion events that cloud AI will filter out than to miss genuine events due to inappropriate motion mask or sensitivity settings.

    Key Points:

    • Motion detection zone should typically be larger than analytical masks 

    • Test the configuration and monitor camera performance 

    • Verify detection at sufficient distances under all lighting conditions (day and night) 

    Step 7: Set Confidence Threshold

    The confidence setting controls detection sensitivity and reduces false positives:

    1. Locate the Confidence field (with information icon)
    2. Enter a value between 20-95 representing the minimum confidence score required (default: 60)
    3. Higher values (80-95) reduce false positives but may miss some legitimate detections
    4. Lower values (30-50) increase sensitivity but may generate more false alerts
    5. This threshold determines when alerts and webhooks will be sent to your systems 

    Step 8: Apply and Save Configuration

    1. Click Apply changes to activate your analytical rule with the specified parameters
    2. Click Save to permanently store your configuration
    3. Use Cancel or Close to exit without applying changes if needed

    Best Practices

    • Test Your Masks: After configuration, monitor initial results to ensure masks are properly positioned and sized for your specific use case.
    • Optimize Loitering Settings: Adjust duration and object count based on your environment—shorter durations for high-security areas, longer for reducing nuisance alerts.
    • Multiple Masks Strategy: Use separate masks with different object types rather than one large mask covering multiple areas, allowing for more granular control.
    • Regular Review: Periodically assess alert patterns and adjust confidence levels to maintain optimal balance between detection accuracy and false positive rates.

    Technical Considerations

    All analytical processing leverages 3dEYE's cloud infrastructure, ensuring consistent performance regardless of camera hardware capabilities. The system supports ONVIF-compliant cameras and integrates seamlessly with webhook configurations for third-party system integration. Analytical rules operate independently per camera, allowing customized detection strategies across your entire deployment.

    Support

    For additional assistance with analytical rule configuration, webhook integration, or troubleshooting detection accuracy issues, contact technical support.

    This feature requires an active 3dEYE subscription with Cloud Analytics enabled. 


        

    Tags
    camera troubleshooting Analytics Advance Analytics Object Detection Mask Loitering Triggers Confidence Threshold Configuration Support
    Published