Tag: AI

Categories
Knowledge Support Support Illustra exacqVision Server Categories Products

Performance Issues Caused by Disabling Metadata while Camera is Connected

Description

Disabling AI metadata within a camera while connected to the exacqVision Server application could lead to camera performance issues.

Product

Any IP Camera providing AI classification metadata.

While this was specifically seen and tested using Illustra devices it can be expected to occur with any camera providing metadata such as AI Object Classification.

Steps to Reproduce

  1. Connect a camera providing AI Object Classification or LPR metadata to the exacqVision Server.
  2. While the device is connected and streaming, enter the camera web browser interface.
  3. Within the camera settings disable the metadata stream. For Illustra Pro Gen4 and Flex Gen4 cameras, this is done by unchecking the ‘Enable AI Object Classification‘ box on the AI Object Classification tab.

Expected Results

Users probably expect no visible change and for the camera to cease streaming metadata.

Actual Results

The camera may become sluggish to respond to commands or provides reduced video performance.

The exacqVision Server logs show repeated entries stating:

Network error 103 in metadata stream

Error code 103 is defined by the exacqVision plugin. The value ‘103’ could change when upgrading or downgrading the exacqVision Server software. The value is calculated as the last libcurl error + 2.

Solution

  1. Navigate to the Add IP Cameras page in the exacqVision Client software.
  2. In the IP Camera List, locate the camera.
  3. Uncheck the box in the Enabled column.
  4. Wait a few seconds, then check the Enabled box to re-enable the connection.
  5. In some cases this may not resolve the issue and the camera will need to be rebooted.

<br>

Categories
Knowledge Support Configuration Support exacqVision Client exacqVision Server Categories Products

Analytic Metadata Best Practices

With Artificial Intelligence being used more often users may notice a spike in the analytic metadata being processed and stored.  Here are our recommendations to manage this increase.

Update exacqVision Server/Client Software

  • Ensure you are running the latest version of both exacqVision Server and Client software.  
  • Search “updating” in the exacqVision Client User Manual, available Here for information on updating.

Adjust Camera Settings

  • Access Analytics from the camera’s configuration GUI page
  • Only those analytic types required by the customer should be enabled
  • The goal is to keep this setting as minimal as possible, while still achieving the customers goals.

Reduce DB retention period

  • Reduce ‘Maximum Days to Keep Bounding Box Metadata’ available in exacqVision Server/Client 23.09 and higher. See: Maximum Days to Keep Bounding Box Metadata Explained
  • Reduce the ‘Maximum Days to Keep Logged Events’
  • These settings decide when older items should be removed or “Purged” from the Database (eventPI), the lower the value, the less storage needed. 
  • It is recommended to set this value as low as possible and adjust based on available storage when data is processing
  • Customer retention requirements must also be considered.

Disable Analytics

  • Each camera is different, but if the user does not have enough storage for 24 hours of analytics, or if analytics are not needed, currently the best option is to disable the analytics completely. 
  • Within the exacqVision Client software, go to Configuration → Server → Configure System → Add IP Cameras → Camera Recording → Analytics Camera → Analytics tab – Uncheck “Record analytic data”

AESW-5987

Categories
Knowledge Support Support exacqVision Server Categories Products

Artificial Intelligence Object Classification Event Triggers Filling OS Drive

Description 

We have found that event links based on triggers from AI Object Classification events flood the eventPI database resulting in the OS drive being filled in a short amount of time. Periodic unexpected crashes, system reboots and increased memory usage have resulted. This happens when the amount of AI metadata captured in the eventPI database in a single day exceeds the available free space on the drive.

The following exacqVision Server log entries confirm when this issue is occurring.

Product 

  • AXIS Object Analytics
  • Hanwha AI-based object detection
  • Illustra AI Object Classification
  • exacqVision Server [All versions]

Solution

Update the exacqVision Server and Client to version 23.09 or higher.

Once updated, utilize the ‘Maximum Days to Keep Bounding Box Metadata’ control. Details on this feature may be found in Maximum Days to Keep Bounding Box Metadata Explained

Related Articles

AES-392