Occupancy determination method and system

A system for determining an actual occupancy distribution of one or more parking spots of a parking area, the system comprising

• a camera with a field of view for capturing images of the parking area, the images comprising representations of the parking spots;
• a processor configured to:
• process an actual image of the parking area to detect one or more vehicles in the actual image, resulting in an actual vehicle representation;
• and
• process the actual vehicle representation and an earlier generated general layout of the parking area to determine the actual occupancy distribution, wherein the general layout is generated based on prior actual images of the area taken by the camera;
• and
• a memory;
• wherein the camera captures a plurality of images over time;
• the processor is further configured to determine an actual vehicle representation for each image of the plurality of images, wherein the vehicle representations are stored in the memory;
• and
• the processor is further configured to process at least a subset of the determined vehicle representations stored in the memory in a spot clustering step to determine the general layout of the parking area and use a probabilistic model to determine the current occupancy distribution based on a comparison of the actual vehicle representation with all possible occupancy distributions of the parking area according to the general layout of the parking area;
• wherein the model processes the actual vehicle representation to calculate a likelihood for each one of the possible occupancy distributions OD(k) with k=1, . . . , 2{circumflex over ( )}J and J being the number of parking spots of the parking area, a respective likelihood expressing to which extent the respective occupancy distribution matches with the actual vehicle distribution;
• or
• the model determines the current occupancy distribution using a calculation of an occupation likelihood P_j for each individual parking spot of interest according to the general layout the parking area, with j=1, . . . , J and with J representing the number of parking spots according to the general layout, a probabilistic estimation based on the actual vehicle representation and the general layout is conducted for each such parking spot of interest, such estimation resulting in likelihoods P_j that such respective parking spot of interest j is occupied by a vehicle, and the actual occupancy distribution is represented by the likelihoods P_1, . . . , P_J.

A method for determining a general layout of a parking area comprising at least one parking spot, the method comprising:

• providing a plurality of vehicle representations VRep(c) with c=1, 2, 3, . . . , wherein different vehicle representations VRep(c1), VRep(c2) with c1/c2 of the plurality of vehicle representations VRep(c) have been determined in corresponding vehicle detection steps from different images of the parking area taken by a camera integrated in a charging station, wherein such different images show the parking area at different points in time;
• and
• processing the vehicle representations in a spot clustering step to determine the general layout of the parking area based on prior actual images of the area taken by the camera;
• wherein the spot clustering applies a hierarchical clustering process on the plurality of vehicle representations;
• processing at least a subset of the determined vehicle representations to determine the general layout of the parking area and use a probabilistic model to determine the current occupancy distribution based on a comparison of the actual vehicle representation with all possible occupancy distributions of the parking area according to the general layout of the parking area;
• wherein the model processes the actual vehicle representation to calculate a likelihood for each one of the possible occupancy distributions OD(k) with k=1, . . . , 2{circumflex over ( )}J and J being the number of parking spots of the parking area, a respective likelihood expressing to which extent the respective occupancy distribution matches with the actual vehicle distribution;
• or
• the model determines the current occupancy distribution using a calculation of an occupation likelihood P_j for each individual parking spot of interest according to the general layout the parking area, with j=1, . . . , J and with J representing the number of parking spots according to the general layout, a probabilistic estimation based on the actual vehicle representation and the general layout is conducted for each such parking spot of interest, such estimation resulting in likelihoods P_j that such respective parking spot of interest j is occupied by a vehicle, and the actual occupancy distribution is represented by the likelihoods P_1, . . . , P_J.