Configuring the License plate recognition XR

Specify the time in seconds after which an event must be generated after recognizing a license plate in the frame. The value must be in the range [0, 60]

Effect of values:

• 0—an event is generated when the vehicle leaves the detection area. The best result is selected from the entire observation period.
• 1 or greater—an event is forced to be generated N seconds after the license plate is first detected, even if the vehicle is still in the frame.

Recommendations:

• 0—the best detection is at the END of the area (vehicle is moving away, rear license plate is visible).
• 1–2—the best detection at the BEGINNING of the area (vehicle is approaching, front license plate is visible).
• Mean value— bidirectional traffic in one lane.

Impact on resources:

• No 

The parameter is disabled by default. To enable the extra processing for license plate search, select the Yes value

Effect of values:

• No—the standard algorithms of license plate recognition are used. Reduces the CPU load.
• Yes—the advanced detection algorithms are used. Improves the detection of non-standard, damaged, or dirty license plates. Increases the CPU load.

Recommendations:

• Enable (the Yes value)—for dirty license plates, non-standard sizes, complex angles, low image contrast.
• Disable (the No value)—high CPU load, standard conditions.

Impact on resources:

• Increased CPU load when this option is enabled

Specify the minimum time in seconds between two returns of the same recognized license plate. The value must be in the range [0, 999].

If the vehicle is within the camera's field of view and the license plate is read correctly, the result is returned at the specified interval. For example, if the interval is 10 seconds, the license plate is returned every 10 seconds, rather than in every frame.

The parameter doesn't affect the frame processing frequency (internal constant ≈ 100 ms).

Effect of values:

• 0 (default value)—no delay. The same license plate can be returned in every frame. Recommended for moving vehicles.
• 1–59 seconds—short delay for moderate filtering.
• 60 seconds or more—optimal for most scenes. Prevents redundant return of license plate from nearby vehicles.

Recommendations:

• 0 seconds—fast traffic flow, short presence in the frame.
• 60–120 seconds—standard use cases.
• 300 seconds—long-term parking.

Impact on resources:

• No 

By default, on the layout and in the archive, the characters of a license plate are completely hidden "GDPR ON". To display the characters of a license plate, you must enter a value:

• "-"—hide all characters (default value),
• " (blank field of a parameter value)"—show all characters,
• "N-"—hide the first N characters,
• "-N"—hide the last N characters,
  where N is the number of characters

Specify the maximum number of readings of the same license plate. The value must be in the range [1, 999]. When the set limit is reached, the result is returned immediately, without waiting for the timeout to expire.

Effect of values:

• Low (1–5): Results are returned faster. Fewer reads are performed to select the best option, potentially resulting in lower accuracy.
• Medium (5–15): Optimal balance between speed and accuracy.
  Recommended for most use cases.
• High (15 or greater): More reads are performed to select the best result. Result latency increases.

Recommendations:

• 5–10—standard moving traffic.
• 10–20 — slow traffic, barriers, checkpoints.

Impact on resources:

• The higher the value of the parameter, the higher the CPU load. 

Specify the minimum value of recognition authenticity as a percentage. The value must be in the range [1, 100]. The parameter determines the minimum average confidence level of all characters of a license plate. 

If the average value of all characters has a confidence level below the set threshold, the entire license plate is rejected and no event is generated

Effect of values:

• 50–70% (low threshold)—more license plates are accepted in general. License plates with individual poorly readable characters can pass filtering.
• 70–85% (medium threshold)—balanced filtering. Recommended for most use cases.
• 85–95% (high threshold)—only license plates with a high degree of confidence are accepted. Recommended for high-quality images and countries with open grammars (for example, the USA).

Difference from the Minimum character confidence threshold parameter

• Minimum authenticity threshold (overall threshold) checks the average confidence value of all characters. Example: character confidence values ​​are [95, 95, 60, 95, 95, 95]. The average value is 89%. The license plate is accepted if the threshold is set below 89%.
• Minimum character confidence threshold (character threshold) checks each character individually. The same license plate is rejected if the threshold is set above 60%, as one character is below the threshold.

Recommendations:

• 50–70%—difficult shooting conditions.
• 70–85%—standard conditions.
• 85–95%—high-quality images, USA (custom license plates).

Impact on resources:

• No

Specify the minimum threshold value of license plate recognition confidence as a percentage. The value must be in the range [1, 100]. The parameter determines the minimum level of recognition confidence for each individual character of the license plate.

If at least one character has a confidence level below the set threshold, the entire license plate is rejected and no event is generated.

Effect of values:

• 50–70% (low threshold)—more license plates are accepted, including those that are difficult to read. The risk of character recognition errors increases (for example, O ↔ 0, B ↔ 8, and so on).
• 70–85% (medium threshold)—balanced filtering. Recommended for most use cases.
• 85–95% (high threshold)—only characters with a high degree of confidence are accepted. The number of errors decreases, but the risk of skipping dirty or damaged license plates increases.

Recommendations: 

• 50–70%—difficult conditions: dirt, damage, poor lighting.
• 70–85%—standard conditions.
• 85–95%—high-quality images, stable lighting.

Impact on resources:

• No

Specify the minimum number of readings of the same license plate.

For example, if you set the value to 2, you will receive a massage about a license plate only if two identical readings have been made. The value must be in the range [1, 999].

Effect of values:

• 1 (the default value)—the license plate is returned after the first reading. Provides maximum coverage, but increases the likelihood of false events.
• 2—the license plate must be read at least twice. Significantly reduces the number of false events. The license plate can be skipped if the vehicle quickly leaves the field of view.
• 3 or greater—high filtering level. Increases the risk of skipping license plates in fast-moving traffic.

Recommendations: 

• 1—fast traffic, skipping license plates is forbidden.
• 2—standard conditions, optimal balance of accuracy and completeness. Particularly recommended for the USA (custom license plates).
• 3 or greater—only for slow traffic, when accuracy is critical.

Impact on resources:

• No 

Specify the period of time in seconds during which repeatedly recognized license plates are ignored. The value must be in the range [0, 86400].

 Effect of values:

• 0 (the default value)—history is disabled. Each recognition generates a separate event.
• 1–59 seconds—short-term filtering. One license plate can generate no more than one event during the specified interval.
• 60–300 seconds—medium filtering. Optimal for checkpoints and areas where vehicles can be stopped.
• 3600 seconds and greater—long-term filtering (hours). Suitable for entrance logs where only the first vehicle appearance is recorded.
• 86400 seconds (maximum)—one license plate, one event per day.

Difference from the Frame processing period (sec) parameter:

• Frame processing period (sec)—filtering at the SDK level, before generation of the final result.
• Period of ignoring repeated recognitions (sec)—filtering of already generated events at the system level.

Recommendations: 

• 0 seconds—parking lots with short stops, each entry and exit must be recorded.
• 60–300 seconds— checkpoints, elimination of duplicates of parked vehicles.
• 3600+ seconds—access control to the area, only the first entry is important.

Impact on resources:

• Minimum memory usage for storing the history of recognized license plates 

Specify the recognition timeout in seconds. The value must be in the range [0, 999]. The parameter determines the period during which multiple readings of the same license plate are accumulated before the final decision is made. After the timeout, the system selects the best result from all accumulated readings.

Effect of values:

• 0 seconds—the result is generated immediately after the first reading. Minimum memory usage, minimum accuracy.
• 1–2 seconds—short accumulation period. Suitable for fast traffic (100 km/h and above).
• 3–5 seconds—standard accumulation period. Optimal balance of speed and accuracy for most use cases.
• 5 seconds or greater—long accumulation period for maximum accuracy. Recommended for slow-moving traffic, areas with barriers, and checkpoints. Increases memory usage.

Interaction with the Maximum/Minimum number of same LP readings parameters

• If the timeout expires but the minimum number of readings has not been reached, the result is discarded.
• If the maximum number of readings is reached before the timeout expires, the result is returned immediately.

 Recommendations:

• 0 seconds—only for static scenes or external triggers.
• 1–2 seconds—highways, fast traffic.
• 3–5 seconds—standard city traffic.
• 5 seconds or greater— slow traffic, barriers, checkpoints.

Impact on resources:

• The higher the value of the parameter, the higher the memory usage

Albania, Algeria, American Samoa, Andorra, Angola, Anguilla, Antigua and Barbuda, Argentina, Armenia, Aruba, Australia, Austria, Azerbaijan, Bahamas, Bahrain, Barbados, Belarus, Belgium, Benin, Bermuda, Bhutan, Bolivia, Bosnia, Botswana, Brazil, Brunei, Bulgaria, Burkina Faso, Burundi, Cambodia, Cameroon, Canada, Cape Verde, Cayman Islands, Central African Republic, Chad, Chile, Christmas Island, Colombia, Comoros, Congo, Cook Islands, Croatia, Cyprus, Czech Republic, Denmark, Dominica, Dominican Republic, East Timor, Ecuador, Egypt, El Salvador, Equatorial Guinea, Estonia, Faroe Islands, Fiji, Finland, France, French Guiana, French Polynesia, French Southern Territories, Gabon, Gambia, Georgia, Germany, Ghana, Gibraltar, Great Britain, Greece, Greenland, Grenada, Guadeloupe, Guatemala, Guinea, Guyana, Haiti, Honduras, Hong Kong, Hungary, Iceland, India, Indonesia, Ireland, Israel, Italy, Ivory Coast, Jamaica, Jordan, Kazakhstan, Kenya, Kiribati, Kosovo, Kuwait, Kyrgyzstan, Laos, Latvia, Liberia, Liechtenstein, Lithuania, Luxembourg, Macau, Macedonia, Madagascar, Malawi, Malaysia, Mali, Malta, Marshall Islands, Martinique, Mauritania, Mauritius, Mayotte, Mexico, Micronesia, Moldova, Monaco, Mongolia, Montenegro, Morocco, Mozambique, Namibia, Nauru, Netherlands, Netherlands Antilles, New Caledonia, New Zealand, Nicaragua, Niger, Nigeria, Northern Mariana Islands, Norway, Oman, Pakistan, Panama, Papua New Guinea, Paraguay, Peru, Philippines, Poland, Portugal, Qatar, Reunion Island, Romania, Russia, Rwanda, Saint Kitts and Nevis, Saint Lucia, Samoa, San Marino, Sao Tome and Principe, Saudi Arabia, Senegal, Serbia, Seychelles, Singapore, Slovakia, Slovenia, Solomon Islands, South Africa, South Sudan, Spain, Sri Lanka, Swaziland, Sweden, Switzerland, Taiwan, Tajikistan, Tanzania, Thailand, Togo, Tonga, Trinidad and Tobago, Turkey, Turkmenistan, Turks and Caicos Islands, Uganda, Ukraine, United Arab Emirates, United Kingdom, United States, Uruguay, Uzbekistan, Vatican, Venezuela, Vietnam, Virgin Islands British, Virgin Islands USA, Wallis and Futuna Islands, Zambia, Zimbabwe, Not used