Skip to content

Reference

Classifiers

Classifiers, short for Few-Shot Learning (FSL) classifiers, can be utilized in queries to filter detections. They are lightweight, requiring only a few dozen examples for training.

Custom Classifiers

If your needs aren't met by the existing classifiers, feel free to contact us by opening a ticket. Custom classifiers can be developed, in binary format, that can integrate seamlessly with other systems, eliminating the need for a new software release.

Classifier Available* Applicable Object Type Classes
Gender yes Person
  • Male
  • Female
Age planned Person 10-year age bands (0-80)
Person on the phone yes Person
  • Talking on the phone
  • Looking at phone
  • None of the above
Hands up yes Person
  • A person with raised hands.
  • A person with non-raised hands.
Emergency vehicle yes Car, Truck
  • Emergency vehicle
  • Non-emergency vehicle
PPE helmet yes Person
  • Wearing
  • Not wearing
PPE vest yes Person
  • Wearing
  • Not wearing
Person wearing a face mask yes Person
  • Wearing
  • Unsure if wears or not.
Person with a gun planned Person
  • With a gun
  • Without a gun
Person is smoking planned Person
  • Smoking
  • Not smoking
Face of person is hidden planned Person
  • Visible face
  • Hidden face
Gate/door open/closed planned Background
  • Open
  • Closed
Simple pose planned Person
  • Standing
  • Sitting
  • Lying down
Fire and smoke planned Background
  • Fire/smoke visible
  • Normal
Construction vehicles planned Car, Truck 12 construction vehicle classes

*If "yes", the classifier is available in the latest release. Otherwise, it is planned for a future release.

Query Types

IRIS+ Professional provides a wide range of use cases across multiple domains. For advanced needs, generic use cases can be leveraged to create custom use cases tailored to their specific needs (via the Generic Object Search, for example). Specialized use cases can be quickly configured by simply filling in a few parameters.

All use cases listed below can be executed in two modes:

  • Forensic Mode: Processes historical data by specifying a start and end time.
  • Live Mode: Runs continuously to generate real-time data or event streams.

All use cases that produce events can also be configured to aggregate events and to produce statistics.

Both event and statistical data can be visualized in the IRIS+ Professional application, and can be forwarded to 3rd party applications such as a VMS.

Many use cases use classifiers, see above, to detect object attributes or actions.

Custom use cases

If you have unique requirements that aren't addressed by existing use cases, feel free to contact us by opening a ticket. Custom use cases can be developed, as binaries, that can integrate seamlessly with your system, eliminating the need to wait for a new software release.

Name Description Main Configuration Parameters Domain
Near Miss Detection Detects if a moving vehicle is close to a person.
  • Vehicle Classes
  • Vehicle Attributes (list of classifiers with desired classes)
  • Distance
 Safety
PPE Detection in Area Detects if a person is wearing a hard helmet and/or vest in a given area.
  • Area Polygon
 Safety
PPE Detection at Entrance Detects if a person is wearing a hard helmet and/or vest when entering an area.
  • Crossline
 Safety
Fall Detection Detects persons in prone position. Safety
Intrusion Detection Detects if a person appears in a protected area.
  • Area Polygon
  • Sensitivity
 Security
Loitering Detection Detects if a person stays in a given area for a configurable amount of time.
  • Area Polygon
  • Max Time Threshold
 Security
Tailgating Detects tailgating for persons or vehicles.
  • Object Class
  • Min Following Time Threshold
 Security
Measure Vehicle Speed* Measures vehicle speed on one camera based on multiple crossing lines.
  • Crosslines
  • Real Distance of Crosslines
 Smart City
Measure Vehicle Following Distance* Measures the following distance of vehicles.
  • Crosslines
 Smart City
Wrong Vehicle* Detects if configured vehicle type is in a given area. Examples include:
  • Only branded vehicles allowed
  • Wrong vehicle in lane (e.g., car in bus lane)
  • Area Polygon
  • Vehicle Attributes (list of classifiers with alert classes)
 Smart City, Security
Single Crossline Traffic Counting Counts line-crossing objects (persons, vehicles, etc.). Counting can be grouped by attributes, direction, and time window.
  • Crossline
  • Attributes (list of classifiers with classes of interest)
  • Aggregation Time Window
 Smart City, Retail
Multiple Crossline Traffic Counting Counts line-crossing objects (persons, vehicles, etc.). Counting can be grouped by attributes and time window. Two crosslines can be added to filter directional traffic.
  • Crosslines and Directions
  • Attributes (list of classifiers with classes of interest)
  • Aggregation Time Window
 Smart City, Retail
Red Light Running Detects if a vehicle runs a red light.
  • Crossline
  • Traffic Light ROI
 Smart City
Measure Queue Length* Measures the number of persons in a queue.
  • Area Polygon
  • Aggregation Window Size
 Retail

* Part of an upcoming release