Attribute Filtering Parameters#
Sometimes, when using multiple classifiers, recognition algorithms may identify conflicting attributes. For example, the system might simultaneously classify a vehicle as both public transport and special equipment.
In such cases, to ensure that the CARS API can correctly interpret the recognition results and determine which attribute to trust, threshold values have been set, and rules for their interpretation have been proposed.
There are two types of thresholds: soft thresholds and hard thresholds.
Results considering the thresholds, as well as their configuration and modification, are presented in the CARS Analytics UI. For detailed information, refer to the section «CARS Analytics. User Guide».
Soft thresholds#
Soft thresholds are used for basic filtering. If conflicting attribute values are detected and the recognition accuracy of one of them is below the specified threshold, it is considered that the value of that attribute is erroneous.
Table 30. Soft thresholds values
| Attribute | Threshold value |
|---|---|
| Vehicle type | 0.8 |
| Vehicle model | 0.6 |
| Emergency type | 0.85 |
| Public transport type | 0.75 |
| Special transport type | 0.8 |
| Vehicle color | 0.92 |
Hard thresholds#
Hard thresholds represent an additional level of filtering, recommended to be used alongside soft thresholds for more accurate interpretation of results. For optimal results, hard thresholds should be applied in conjunction with the rules described below.
Table 31. Hard thresholds values
| Attribute | Threshold value |
|---|---|
| Vehicle type | 0.9 |
| Vehicle model | 0.75 |
| Emergency type | 0.9 |
| Public transport type | 0.9 |
| Special transport type | 0.9 |
| Vehicle color | 0.92 |
The rules for filtering attributes using hard thresholds are presented in Tables 32 and 33.
Table 32. Filtering Rules 1-4 with Hard Thresholds
| Condition 1 | Operator | Condition 2 | Result |
|---|---|---|---|
| Rule 1 | |||
| Attribute «Public transport type» != «Other» | AND | Evaluation of recognition accuracy > hard threshold for the type of public transport | Attribute «Emergency Service Type» == «Non-Emergency» (regardless of other ratings) and attribute «Special Transport Type» == «Non-special» (regardless of other ratings) |
| Rule 2 | |||
| Attribute «Vehicle type» == A | AND | Evaluation of recognition accuracy > hard threshold for vehicle type | Remaining attributes with thresholds (that have defined thresholds) are not relevant |
| Rule 3 | |||
| Attribute «Vehicle type» == B_light | AND | Evaluation of recognition accuracy > hard threshold for vehicle type | Attribute «Special Transport Type» == «Non-special» (regardless of other ratings) |
| Rule 4 | |||
| Attribute «Vehicle type» == B_heavy | AND | Evaluation of recognition accuracy > hard threshold for vehicle type | Allowed values for «Special Transport Type» are «Non-special» or «Truck», otherwise «Undefined» |
Table 33. Filtering Rule 5 with Hard Thresholds
| Condition 1 | Operator | Condition 2 | Operator | Condition 3 | Result |
|---|---|---|---|---|---|
| Rule 5 | |||||
| Evaluation of the recognition accuracy of the vehicle model < hard threshold for the vehicle model | AND | Estimation of vehicle type recognition accuracy < hard threshold for vehicle type | AND | Recognition accuracy for emergency transport type < hard threshold | The remaining attributes (that have defined thresholds) are not relevant |