Skip to main content

ANSVIS v1.2.0.26

Face Recognition

  • Update detection behavior: When user configure the task to only trigger one event per ID, the unknown ID will still be triggered again if the face is recognized. With this update, the system will now trigger an initial event when the face is unknown and a second event when it becomes known, ensuring that recognized faces are always captured even if they were previously marked as unknown.
  • New users are now required to have a user ID. If the field is left empty during creation, the system will automatically generate an ID for the user, which can be modified later if needed.
  • Add face score parameter to control face detection confident score
  • Add face liveness and anstispoofing information for detected face
  • Update face labelling process in face recording task
  • Face Recording Task: Now user can pause the recording to view images.
  • Support import/export face database to quickly register faces
  • Enhance detection accuracy and processing speed

Performance

  • Add Active Engine setting to control the number of AI tasks running in parallel for resource optimization.
  • Add Input Trigger to create task execution condition. This feature allows automated event capture and processing based on external signals, enhancing flexibility for access control, traffic management, and security applications (available for pre-built and custom tasks)
  • System-Wide Performance Improvements: Overall system responsiveness and resource utilization have been optimized, resulting in lower CPU/GPU usage, reduced memory footprint, and improved throughput for high-density camera deployments.
  • Improve ALPR performance with faster detection speeds and higher accuracy rates
  • Fix an issue where cameras continued streaming in the background after a task was stopped. With this fix, camera streams will now properly terminate when the associated task is stopped, ensuring optimal resource usage and system performance.

Rules

  • Add new rule join condition IF-THEN
  • Fix path rule behavior not detecting object in some movement patterns.
  • Add Line ROI Gap to increase line thickness, reducing false detection when object wanders around the line (available for pre-built and custom tasks)

Triggers

  • Fix incorrect format of custom HTTP trigger data when object has no attribute.
  • Add Fully Serialized custom format that split all detected objects in an event into separated HTTP trigger.
  • The HTTP Send Test function now is done from the ANSVIS Server instead of ANSVIS Client.
  • Email trigger: Fixed an issue where the attached image did not include bounding boxes.

Pre-built AI Tasks

  • Update description for each pre-built AI Task. Click the i icon next to task name to view description.
  • Update configuration layout
  • Add model sensitivity and dedicated smoke score for Fire & Smoke Detection task
  • Add Improper Object Placement task: Detects objects (e.g., such as vehicles, goods, waste, and other items) that are placed in areas where they should not be, causing obstruction or violating spatial rules.

Others

  • Add ANS Objects model that include some objects to monitor
  • Update definition for model score and model sensitivity
  • Image snapshot will take a few more seconds to ensure good image being captured
  • Add Operation Note to ensure ANSVIS Server operate as expected.
  • The AIBOX mobile app is now available worldwide on both the App Store and Google Play