This is Person Identification Test App using Intel OpenVINO Person Re-Identification Model.
You can do followings:
- Person Detection
- Preson Re-Identification (Tracking and Counter)
Person re-identifiction - Tracking - (YouTube Link)
Person re-identifiction - Tracking and Counter- (YouTube Link)
- Python 3.7.6 (need 3.6+ for f-strings)
- Windows 10 [Version 10.0.19041.388]
- OpenVINO Toolkit 2020.1+1
pip install -r requirements.txt
python app.py -h
usage: app.py [-h] -i INPUT [-d {CPU,GPU,FPGA,MYRIAD}]
[-d_reid {CPU,GPU,FPGA,MYRIAD}] [--v4l] [-ax {0,1}] [-g GRID]
optional arguments:
-h, --help show this help message and exit
-i INPUT, --input INPUT
Path to video file or image. 'cam' for capturing video
stream from camera
-d {CPU,GPU,FPGA,MYRIAD}, --device {CPU,GPU,FPGA,MYRIAD}
Specify the target device for Person Detection to
infer on; CPU, GPU, FPGA or MYRIAD is acceptable.
-d_reid {CPU,GPU,FPGA,MYRIAD}, --device_reidentification {CPU,GPU,FPGA,MYRIAD}
Specify the target device for person re-identificaiton
to infer on; CPU, GPU, FPGA or MYRIAD is acceptable.
--v4l cv2.VideoCapture with cv2.CAP_V4L
-ax {0,1}, --axis {0,1}
Specify the axis when counting person horizontally or
vertically (0: count horizontally(x-axis) , 1: count
vertically (y-axis)
-g GRID, --grid GRID Specify how many grid to divide frame. This is used to
define boundary area when the tracker counts person. 0
~ 2 means not counting person. (range: 3 < max_grid)
example1. camera streaming without person counter
python app.py -i cam
example2 Specify video file with person counter
python app.py -i P1033674_1920.mp4 --grid 10
Access the url bellow on your browser
http://127.0.0.1:5000/
Footnotes
-
openvino.inference_engine version openvino_2020.1.033 or above build does not need cpu extension.
https://software.intel.com/en-us/forums/intel-distribution-of-openvino-toolkit/topic/848825
โฉ