Tracking๐
Capabilities of the module๐
The Tracking module allows to track objects:
crossing the set line in one or both directions;
entering the set zone of arbitrary shape;
being in the zone beyond the specified time.
Depending on the settings, the module can track:
any moving objects;
objects of certain categories:
People
Animals
Vehicles by type: Passenger cars, Buses, Trucks, Motorcycles.
Moreover, it is possible to set up the tracking of objects of a certain range of dimensions.
In real-time viewing mode, controlled lines and zones can be displayed, along with object trajectories, while the objects may be highlighted with frames of different colors.
All the module events are recording in the Events log.
Module operation modes๐
The module can operate in two object tracking modes: Only those in motion and Filter by categories. The object tracking mode Only those in motion can work with or without the use of neural networks. The Filter by categories mode works using neural networks.
Objects being tracked (setting in Configurator) |
Only those in motion |
Filter by categories |
|
Additional setting in the Configurator |
Using neural networks |
||
Available in version 4.3 |
โ |
โ |
โ |
Available in version 4.4 |
โ |
โ |
โ |
Neural networks are used to detect objects |
โ |
โ |
โ |
Neural networks are used to make a trajectory |
โ |
โ |
โ |
Can operate efficiently without a graphics card (GPU) |
โ |
โ |
โ |
It is possible to run the module on a high-performance graphics card (see hardware resource requirements) |
โ |
โ |
โ |
Relative accuracy of trajectories |
Middle |
Average or above average in certain conditions |
High |
Operating systems |
Available on all operating systems used by the Eocortex software |
The list of supported operating systems depends on the :ref:` used neural network package <tracking-requirements-os>` |
|
What objects are tracked |
All moving objects detected by software motion detector. |
Objects of the following categories:
|
Note
If the Only those in motion mode was configured in version 4.3, then after updating to version 4.4 the Only those in motion mode will be set without using neural networks (the Using neural networks option will be turned off).
Details๐
Warning
Starting with version 3.5 of Eocortex, the Tracking, Search for Objects and Uniform Detection modules running on the same camera share the processor and video card resources. This reduces the load on the server.
For the most efficient sharing of resources, it is recommended to set the same operating mode for the camera in the settings of these modules: either using a video card or without using a video card.
The approximate accuracy of determining the types of transport:
Passenger cars ~95%;
Buses ~80%;
Trucks ~70%;
Motorcycles ~90%.
Compatibility with other modules๐
Compatible with modules:
Auto Zoom
Loud sound detection
Fire and Smoke Detection
Face Mask Detection
Sabotage Detection
Crowd Monitoring
Personnel Activity Monitoring
Shelf Fullness Check
Uniform Detection
Face detection
Abandoned object detection module
Fall Detection
Emergency Vehicle Detection
Counting People in Queue
People Counting
Unique Visitor Counting
Search for Objects
Frame Area Blurring
License Plate Recognition (Complete)
License Plate Recognition (Light)
Face Recognition (Complete)
Face Recognition (Light)
Traffic density heat map
Not recommended for use with modules:
Fisheye dewarping module
Incompatible with modules:
Object Classification and Counting
Usage๐
Use of the module in the Eocortex Client application.