Systems, devices, and methods for pedestrian traffic assessment
AssigneePivot Analytics, LLC
InventorsWilliam Royle, Wayne Mitzen, David Dochter, Matthew Alexander
Life of this patent
Estimating the term…
Abstract
Systems, devices, and methods for pedestrian footfall counting are provided. The system includes a plurality of pedestrian footfall counting devices configured to capture and record optical data of pedestrians. The system further includes one or more relay servers configured to receive the optical data from the pedestrian footfall counting devices. The system further includes one or more counting servers configured to process the optical data and determine a pedestrian footfall count based on the optical data.
Claims11
A computer implemented method of quantifying pedestrian footfall, the method to be carried out by a counting server including a CPU core with multithreading capability operating a plurality of CPU threads, and a GPU, the method comprising:
- obtaining by a first optical processing thread of the plurality of CPU threads a first optical data block;
- obtaining by a second optical processing thread of the plurality of CPU threads a second optical data block;
- determining, by the first optical processing thread, a scene change in the first optical data block;
- determining, by the second optical processing thread, a second scene change in the second optical data block;
- requesting, by the first optical processing thread, that a GPU management thread of the plurality of CPU threads allocate GPU resources for analysis of a first optical frame associated with the scene change;
- requesting, by second optical processing thread, that the GPU management thread allocate GPU resources for analysis of a second optical frame associated with the second scene change;
- allocating, by the GPU management thread, first GPU resources for analysis of the first optical frame and second GPU resources for analysis of the second optical frame;
- identifying, by the GPU, a feature of interest in the first optical frame;
- identifying, by the GPU, a second feature of interest in the second optical frame;
- tracking, by the first optical processing thread, the feature of interest in optical frames of the first optical data block subsequent to the first optical frame;
- tracking, by the second optical processing thread, the second feature of interest in optical frames of the second optical data block subsequent to the second optical frame;
- incrementing, by the first optical processing thread, an FOI count responsive to the feature of interest crossing a predetermined point in one of the optical frames;
- and
- incrementing, by the second optical processing thread, the FOI count responsive to the second feature of interest crossing the predetermined point in a second one of the optical frames.
A pedestrian footfall quantification system comprising:
- a plurality of pedestrian footfall recording devices configured to:
- obtain optical data and transmit the optical data;
- a relay server including at least one CPU configured to:
- receive the optical data transmitted by the plurality of pedestrian footfall recording devices;
- store the optical data in a storage device;
- a counting server, including at least a multi-threaded CPU operating a plurality of CPU threads, and a GPU and being configured to:
- obtain the optical data from the storage device;
- process the optical data to obtain a count of features of interest in the optical data,
- wherein processing the optical data includes:
- requesting, by a first optical processing thread of the plurality of CPU threads, that a GPU management thread of the plurality of CPU threads allocate GPU resources for analysis of a first optical frame,
- requesting, by a second optical processing thread of the plurality of CPU threads, that the GPU management thread allocate GPU resources for analysis of a second optical frame,
- allocating, by the GPU management thread, first GPU resources for analysis of the first optical frame and second GPU resources for analysis of the second optical frame,
- identifying, by the GPU, a feature of interest in the first optical frame, and
- identifying, by the GPU, a second feature of interest in the second optical frame.
Dependencies are inferred from the claim text.