| Attribute |
Vehicle or LP parameter: brand, model, color, LP symbols, etc. |
| Base64 |
A standard for encoding binary data (in the context of the CARS API - images) using 64 ASCII characters |
| Best shot |
Frame of the video stream, on which the object is fixed in the optimal view for further use in the CARS API to determine attributes |
| Bounding box (BBox) |
A rectangle that bounds the image space with a detected object (Vehicle, LP, pedestrian, animal) |
| CARS Analytics |
Subsystem for collecting, storing and displaying the results of the CARS API and CARS Stream operation through a web interface |
| CARS API |
Subsystem for detection and recognition of vehicles and LPs attributes on images in real-time |
| CARS Stream |
Subsystem for detecting and tracking objects in the video stream and determining the best frames |
| Classifier |
An object of the CARS API subsystem that recognizes one of the attributes of a vehicle or LP |
| Crop |
A BBox cropped out of a frame with a detected object (vehicle, LP, pedestrian, animal) |
| Detector |
Algorithm for finding areas of the image containing the object |
| Detection |
Actions to determine areas of the image containing the object |
| Exchangeable Image File Format (EXIF) |
A format that allows a user to add additional information (metadata) to images, commenting on this file, describing the conditions and methods of obtaining it, authorship, etc. |
| False negative detection |
Result of the detection when an object of interest is not detected and not defined as an object of interest |
| False positive detection |
Result of the detection when not an object of interest is detected and defined as an object of interest |
| Foreground subtraction (FGS) |
A technology that subtracts the background of an image and processes only the region with moving objects. It is used to improve the performance of the system |
| JavaScript Object Notation (JSON) |
A text-based data exchange format based on JavaScript |
| Inference |
Image processing by a neural network and generation of results: coordinates of detections and the probability that the detection is an object of interest |
| Intersection Over Union (IOU) |
Parameter determines the coefficient of intersection of two detections |
| License plate (LP) |
An individual alphanumeric designation assigned to a vehicle by a registration unit |
| LUNA CARS |
A system designed for detection, tracking of objects and recognition of vehicle and LPS attributes. The system consists of three subsystems: CARS Analytics, CARS API and CARS Stream |
| Nginx |
A load balancer that allows you to distribute requests and responses between running copies of the CARS API and external systems |
| Non-maximum suppression |
Algorithm for determining the boundaries of objects in the image |
| Object |
An item of interest for the system within the detection, re-detection and tracking processes: vehicle, LP, smoke, fire, pedestrian or animal |
| Redetection |
Repeated action to find areas of the image containing the object. Produced in the area of the image found using detection |
| Reidentification |
Repeated action to identify an object in order to «glue» the broken tracks of the same object into one |
| Track |
Information about the position of the object in the sequence of frames |
| Tracking |
The process of obtaining information about the position of the object on a sequence of frames |
| Transmission Control Protocol (TCP) |
A real-time streaming protocol that contains a simple set of basic commands for controlling the video stream |
| Real Time Streaming Protocol (RTSP) |
A transport-layer data transfer protocol in connectionless networks that provides the delivery of datagrams without requiring confirmation of their receipt |
| Vehicle |
A technical device for the transport of people and/or goods, for which a special right is granted |
| User Datagram Protocol (UDP) |
A transport-layer data transfer protocol in connectionless networks that provides the delivery of datagrams without requiring confirmation of their receipt |