Now you can specify the recognition scenario for post-analytic. The main goal is to provide one recognition event per vehicle without duplicates.
The "Default" scenario covers 90% of real-life cases (traffic monitoring in the highways/city streets, parking, etc).
This scenario has complicated logic which also covers the "Traffic flow" scenarios below.We added more scenarios for some "hard" installation cases:
"Parking entry" - this scenario will help you in cases when a vehicle approaches the camera at low speed.
"Stationary car" - this scenario could help you in cases when the car is parked in front of the camera and stays for a long time (from several minutes to several hours).
"Test mode" - could be useful while checking recognition quality. This mode disables post analytics and in the results tab, you will get all raw recognition results.
"Traffic flow - best result on the confidence" - this is a private work scenario. For example, if a car license plate was recognized only 10 times, then the algorithm from 10 recognition results will select one result with the highest confidentiality for display.
"Traffic flow - best result on the statistic" - This is a private work scenario. For example, if a car license plate was recognized only 10 times, while 7 times the number was recognized incorrectly with a confidentiality of 0.6 and three times correctly with a confidentiality of 0.9, then the algorithm, based on the statistics of the number of recognitions will display one result out of 7 incorrect ones since there were more results.