| General parameters |
|
|
|
|
| logSeverity |
The parameter sets the logging mode of the VehicleEngine. |
- 4 – logging debug information; |
3 |
 |
|
|
- 3 – logging information about work plus information for 4; |
|
|
|
|
- 2 – logging warning record plus information for 3; |
|
|
|
|
- 1 – logging error record plus information for 2; |
|
|
|
|
- 0 – disable logging |
|
|
| profiling |
Profiling |
- 0 – profiling is disabled; |
0 |
|
|
|
- 1 – profiling is enabled |
|
|
| defaultVehicle DetectorType |
Vehicle detector by default. |
VehicleDetectorV4 |
VehicleDetectorV4 |
|
| defaultPlate DetectorType |
LP detector by default |
PlateDetectorV5 |
PlateDetectorV5 |
|
| defaultAnimal DetectorType |
Animals detctor by default |
AnimalDetectorV1 |
AnimalDetectorV1 |
|
| defaultSmokeFire DetectorType |
Smoke and Fire detector |
SmokeFireDetectorV1 |
SmokeFireDetectorV1 |
|
| defaultAggregatedPlateAttributesEstimatorType |
A version of the algorithm for estimating the accuracy of recognized LP attributes. Each version of the algorithm contains a specific set of countries to evaluate. For more information, see the configuration files plate_aggregated_v1.conf and plate_aggregated_v2.conf in the CARS API |
V4; V5 |
V5 |
|
| defaultCountryPlateEstimatorType |
A version of the algorithm for evaluating the accuracy of recognizing the country of ownership of the LP. Each version of the algorithm contains a specific set of countries to evaluate. For more information, see the configuration files grz_country_recognition_*.conf |
- CIS_ATTRS_V1; |
V3 |
|
|
|
- CIS_ATTRS_V2; |
|
|
|
|
- V3 |
|
|
| Detector VehicleDetectorV4 |
|
|
|
|
| ScoreThreshold |
Vehicle detection accuracy threshold. Increasing the threshold reduces the number of false positive detections but increases the number of false negative detection (misses). |
0.000…1.000 |
0.50 |
|
| imageSize |
The size of the image in pixels on the larger side on which the vehicle is detected. Increasing the size will allow better detection of vehicles in the background, but this will increase the processing time. If it is necessary to detect a large number of small objects, it is recommended to use the VehicleDetectorV2Large detector. |
320...1280 |
640 |
 |
| minInputSize |
Minimum image size for inference. Measured in pixels. The parameter value must be strictly less than the imageSize value. |
This parameter is selected theoretically, has one default value and must not be changed by users. |
- |
40 |
| squareFormat |
Enable Square Format - Automatically adds black bars to the edges of a rectangular frame to create a square frame. Turning it on will increase image processing time. |
- 0 – on; |
0 |
|
|
|
- 1 – off. |
|
|
| alignment |
If squareFormat = 0 then if possible width or height of the input image will be aligned up to this value, extra space will be filled with black pixels. This parameter is needed for better batching utilization. The parameter value should be 2 to the nth degree (i.e. 2/4/8/16/...). |
2...512 |
64 |
|
| maxFlowerBatchSize |
Maximum number of images used for inference at once. |
1...16 |
16 |
|
| NMSThreshold |
Threshold value for non-maximum suppression. The higher the threshold, the higher the probability of detecting a vehicle blocked by objects. Increasing the threshold will increase the number of false positive detections. |
0.000…1.000 |
0.4 |
 |
| SecondNMSThreshold |
Rechecking the non-maximum suppression threshold, with a larger value. Increasing the value will reduce the number of false positive detections. |
0.000…1.000 |
1.0 |
|
| minSize |
Minimal size of detection. Detections of a smaller size will not be returned to the user and will not be used for the further processing. This parameter is selected theoretically, has one default value and must not be changed by users.й |
- |
20 |
|
| nms |
Parameter affecting the accuracy of vehicle localization on the frame. Higher accuracy, reduces performance. |
- best – the mode in which the best BBox is the one whose intersection with other BBoxes has the maximum value; |
mean |
|
|
|
- mean – the mode in which the best BBox is determined by averaging the values of all intersecting BBoxes |
|
|
| redetectScoreThreshold |
Threshold for estimation of vehicle redetection accuracy. Increasing the threshold reduces the number of false negative detection results. |
0.000…1.000 |
0.25 |
|
| redetectImageSize |
The size of the image in pixels along the larger side, on which the vehicle is redetected. |
100...1280 |
150 |
|
| redetectOneNMSThreshold |
The value of the non-maximum suppression threshold for vehicle redetection. The higher the threshold, the higher the probability of repeated detection of a vehicle blocked by objects. Increasing the threshold will increase the number of false positive detections. |
0.000…1.000 |
0.4 |
|
| redetectNMSThreshold |
The value of the non-maximum suppression threshold for vehicle redetection. The higher the threshold, the higher the probability of repeated detection of a vehicle blocked by objects. Increasing the threshold will increase the number of false positive detections. |
0.000…1.000 |
0.5 |
|
| redetectExpandCoeff |
Image scaling factor.. |
1.000…2.000 |
1.4375 |
|
| redetectSquareThreshold |
The threshold value for skipping redetections by area. Redetections whose area is less than the specified threshold are not used for further work. The parameter value is derived theoretically and cannot be changed |
- |
3.25 |
|
| redetectNMS |
Parameter affecting the accuracy of vehicle localization on the frame. Higher accuracy, reduces performance. |
- best – the mode in which the best BBox is the one whose intersection with other BBoxes has the maximum value; |
mean |
|
|
|
- mean – the mode in which the best BBox is determined by averaging the values of all intersecting BBoxes |
|
|
| Detector PlateDetectorV5 |
|
|
|
|
| ScoreThreshold |
LP detection accuracy threshold.Increasing the threshold reduces the number of false positive detections but increases the number of false negative detection (misses). |
0.000…1.000 |
0.50 |
|
| imageSize |
The size of the image in pixels along the larger side, on which the LP is detected. Increasing the size will allow better detection of small LPs on larger vehicles, but it will increase the processing time. |
320...1280 |
320 |
|
| squareFormat |
Enable Square Format - Automatically adds black bars to the edges of a rectangular frame to create a square frame. Turning it on will increase image processing time. |
- 0 – on; |
1 |
|
|
|
- 1 – off. |
|
|
| maxFlowerBatchSize |
Maximum number of images used for inference at once. |
1...16 |
16 |
|
| NMSThreshold |
Threshold value for non-maximum suppression. The higher the threshold, the higher the probability of detecting a LP blocked by objects. Increasing the threshold will increase the number of false positive detections. |
0.000…1.000 |
0.35 |
|
| SecondNMSThreshold |
Rechecking the non-maximum suppression threshold, with a larger value. Increasing the value will reduce the number of false positive detections. |
0.000…1.000 |
1.0 |
|
| nms |
A parameter that affects the accuracy of localization of the LP on the frame. The higher the accuracy, the lower the speed. |
- best – the mode in which the best BBox is the one whose intersection with other BBoxes has the maximum value; |
best |
|
|
|
- mean – the mode in which the best BBox is determined by averaging the values of all intersecting BBoxes |
|
|
| Detector SmokeFireDetectorV1 |
|
|
|
|
| ScoreThreshold |
Threshold for estimating the accuracy of smoke and fire detection. Increasing the threshold decreases the number of false positives, but increases the number of false negatives (smoke and fire misses) |
0.000…1.000 |
0.50 |
|
| Detector AnimalDetectorV1 |
|
|
|
|
| ScoreThreshold |
Animal detection accuracy threshold. Increasing the threshold reduces the number of false positive detections but increases the number of false negative detection (misses). |
0.000…1.000 |
0.65 |
|
| NMSThreshold |
Threshold value for non-maximum suppression. The higher the threshold, the higher the probability of detecting an animal blocked by other objects. Increasing the threshold will increase the number of false positive detections. |
0.000…1.000 |
0.45 |
|
| imageSize |
The size of the image in pixels along the larger side, on which the animal is detected. Increasing the size will allow better detection of small animals, but it will increase the processing time. |
320...1280 |
512 |
|
| nms |
A parameter that affects the accuracy of localization of the animal on the frame. The higher the accuracy, the lower the speed. |
- best – the mode in which the best BBox is the one whose intersection with other BBoxes has the maximum value; |
mean |
|
|
|
- mean – mode in which the best BBox is determined by averaging the values of all |
|
|
| minSize |
Minimal size of detection. Detections of a smaller size will not be returned to the user and will not be used for the further processing. This parameter is selected theoretically, has one default value and must not be changed by users.й |
- |
16 |
|