Service system and method for detecting number of in-vehicle persons of vehicle in high occupancy vehicle lane

A service method for detecting in-vehicle persons of a vehicle running on a high occupancy vehicle lane in which a control device communicates with a first camera unit disposed diagonally to a traveling direction of a vehicle running on the high occupancy vehicle lane and configured to transmit first image information composed of a plurality of frame images by continuously photographing the vehicle and a second camera unit disposed orthogonally to the traveling direction of the vehicle and configured to transmit second image information composed of a plurality of frame images by continuously photographing the vehicle, the service method comprising:

• detecting an object corresponding to an in-vehicle person of the vehicle for each of a plurality of seat regions respectively corresponding to a plurality of seats, which is separated for the in-vehicle persons sit, by performing image analysis on each of the plurality of frame images constituting the first image information, and then labeling each of seat regions in which an object is detected with a preset seat number corresponding to the seat region, thereby creating first labeling information including one or more frame images labeled with seat numbers;
• creating second labeling information based on the second image information in the way of creating the first labeling information based on the first image information;
• and
• totaling the number of seat numbers such that seat numbers are not repeatedly totaled based on the first labeling information and the second labeling information, and calculate a final number of in-vehicle persons as the totaled number of seat numbers, and determining whether the vehicle violates a law about use of a high occupancy vehicle lane based on the final number of in-vehicle persons,
• wherein the determining whether the vehicle violates a law about use of a high occupancy vehicle lane further includes:
• extracting feature information of an object detected in a seat region for each of seat regions in which an object is detected based on the first image information;
• creating in-vehicle person information including the seat information of the object and a preset seat number corresponding to the seat region in which the object is detected, and matching and storing the feature information with the first labeling information created to correspond to the first image information;
• creating one or more items of in-vehicle person information in the same way as creating the in-vehicle person information based on the first image information, and matching and storing the one or more items of in-vehicle person information with the second labeling information;
• comparing feature information of one or more items of in-vehicle person information matched to the first labeling information or the second labeling information in accordance with a preset similarity algorithm, and when there is a plurality of items of in-vehicle person information in which similarity of feature information is a preset reference value or more and that has different seat numbers, determining the in-vehicle person information as repeated in-vehicle person information;
• setting and deleting the other in-vehicle person information excluding in-vehicle person information corresponding to a most repeated seat number as deletion target in-vehicle person information by comparing the plurality of items of repeated in-vehicle person information;
• checking one or more seat numbers based on one or more items of in-vehicle person information matched to the first labeling information or the second labeling information after the deletion target in-vehicle person information is deleted;
• and
• totaling the number of seat numbers such that same numbers are not repeatedly totaled for the checked one or more seat numbers, and calculating the totaled number of seat numbers as the final number of in-vehicle persons.