Skip to content

Architecture#

The CARS_Stream architecture (Figure 1).

CARS_Stream architecture
Figure 1. CARS_Stream architecture

A detailed description subsystem components is given in Table 2.

Table 2. Description of CARS_Stream components

Component Description
Source Video file or RTSP stream containing areas with vehicle, LP, smoke, fire, pedestrian, animal on the frame
Load balancer A service for distributing requests and responses between several running copies of CARS_Stream and an external analytics system
CARS_Stream Server Service for interaction with external systems and processing of incoming video streams and video files - converts them into a sequence of frames for subsequent detection and tracking of the position of the object and determine the best frame
Track Engine A library containing a set of algorithms for detecting, tracking the position and determining the best frames of the objects
External analytics system An external system that allows aggregating information about the position of the object. Currently, only CARS_Analytics is supported

CARS_Stream operation sequence diagram#

Sequence diagram of processing input data and selecting the best frame (Figure 2).

Sequence diagram of processing input data and selecting the best frame
Figure 2. Sequence diagram of processing input data and selecting the best frame

A detailed description of the diagram is given in Table 3.

Table 3. Sequence diagram description

Description
(1) Video files or RTSP streams are sent to CARS_Stream Server from source
(2) CARS_Stream Server divides video into frames
(3) CARS_Stream Server sends a request to the Track Engine for frame processing
(4) Track Engine detects objects on the frame, generates an object track, determines the start track shot and best shots of object’s track
(5) The Track Engine returns to the CARS_Stream Server information about the position of the object, the object track, start track shot and the best shot
(6) CARS_Stream Server sends a request with the best frames and information about the position of the object to an external system for subsequent retrieval of the object’s attributes