Skip to content

Latest commit

 

History

History
131 lines (100 loc) · 5.5 KB

README.md

File metadata and controls

131 lines (100 loc) · 5.5 KB

Smart Library Demo

The applicattion is a demo of automated “smart library”. It involves the registration of the reader; authorization of the reader through face recognition; receiving and returning books by recognizing QR codes generated for each book in the library. The following pretrained models can be used:

  • face-detection-retail-0004, to detect faces and predict their bounding boxes;
  • landmarks-regression-retail-0009, to predict face keypoints;
  • face-reidentification-retail-0095, to recognize readers.

For more information about the pre-trained models, refer to the model documentation.

How it works

The application is started from command line. It accepted several parameters. It reads video stream frame-by-frame from a web-camera device and performs independent analysis of each frame. To make predictions the application using 3 models. An input frame is processed by the face detection model to predict face bounding boxes. Then, face keypoints are predicted by the facial landmarks regression model. Keypoints are using to align the face and match it with face in data base. To register in library press r on keyboard. Once reader has beeen registered, he will be autothorized through face recognition. Registration allows to receiving and ruturning books by recognizing QR codes for each book. To make book recognition press b on keyboard. Also applications provides some extra statictics in console, like list of registered readers, full list of books in the library, hystory of borrowing books. To change information in console press f on keyboard. To exit press q.

Creating QR-codes for books

Next two files are using to create QR-codes for books: library.json file contains information about books in the library. createQRCodes.py script generates QR-codes for each book in library.json file.

usage: createQRCodes.py [-h] [-i LIB] [-o OUT]

optional arguments: -h, --help show this help message and exit -i LIB unput .json file with info -o OUT directory to save generated QR-codes

Installation and dependencies

The demo depends on:

  • OpenVINO toolkit (2019R3 or newer)
  • Python (any of 2.7+ or 3.4+, which is supported by OpenVINO)
  • OpenCV (>=3.4.0)

To install all the required Python modules you can use:

pip install -r requirements.txt

Running the demo:

Running the application with the -h option or without any arguments yields the following message:

python ./smart_library_demo.py -h

usage: smart_library_demo.py [-h] -reid RDDET -m_rd RDMODEL -fd FDDET -m_fd
                             FDMODEL -lm LMDET -m_lm LMMODEL [-w_rd RDWIDTH]
                             [-h_rd RDHEIGHT] [-t_rd RDTHRESHOLD]
                             [-w_fd FDWIDTH] [-h_fd FDHEIGHT]
                             [-t_fd FDTHRESHOLD] [-w_lm LMWIDTH]
                             [-h_lm LMHEIGHT] [-br BR] [-lib LIB] [-w WEB]

Smart Library Sample

Optional arguments:
  -h, --help         show this help message and exit
  -w_rd RDWIDTH      Optional. Image width to resize
  -h_rd RDHEIGHT     Optional. Image height to resize
  -t_rd RDTHRESHOLD  Optional. Probability threshold for face detections.
  -w_fd FDWIDTH      Optional. Image width to resize
  -h_fd FDHEIGHT     Optional. Image height to resize
  -t_fd FDTHRESHOLD  Optional. Probability threshold for face detections.
  -w_lm LMWIDTH      Optional. Image width to resize
  -h_lm LMHEIGHT     Optional. Image height to resize
  -br BR             Optional. Type - QR
  -lib LIB           Optional. Path to library.
  -w WEB             Optional. Specify index of web-camera to open. Default is
                     0

Models:
  -reid RDDET        Required. Type of recognizer. Available DNN face
                     recognizer - DNNfr
  -m_rd RDMODEL      Required. Path to .xml file
  -fd FDDET          Required. Type of detector. Available DNN face detector -
                     DNNfd
  -m_fd FDMODEL      Required. Path to .xml file
  -lm LMDET          Required. Type of detector. Available DNN landmarks
                     regression - DNNlm
  -m_lm LMMODEL      Required. Path to .xml file


Example of a valid command line to run the application:

Linux (`sh`, `bash`, ...) (assuming OpenVINO installed in `/opt/intel/openvino`):

``` sh
# Set up the environment
source /opt/intel/openvino/bin/setupvars.sh

python ./smart_library_demo.py \
-reid='DNNfr' \
-m_rd=<path_to_model>/face-reidentification-retail-0095.xml \
-fd='DNNfd' -m_fd=<path_to_model>/face-detection-retail-0004.xml  \
-lm='DNNlm' \
-m_lm=<path_to_model>/landmarks-regression-retail-0009.xml \

Windows (cmd, powershell) (assuming OpenVINO installed in C:/Program Files (x86)/IntelSWTools/openvino/):

# Set up the environment
call C:/Program Files (x86)/IntelSWTools/openvino_2019.3.334/bin/setupvars.bat

python smart_library_demo.py  -reid='DNNfr' -m_rd=<path_to_model>/face-reidentification-retail-0095.xml 
                              -fd='DNNfd' -m_fd=<path_to_model>/face-detection-retail-0004.xml 
                              -lm='DNNlm' -m_lm=<path_to_model>/landmarks-regression-retail-0009.xml

Notice that the custom networks should be converted to the
Inference Engine format (*.xml + *bin) first. To do this use the
[Model Optimizer](https://software.intel.com/en-us/articles/OpenVINO-ModelOptimizer) tool.

### Demo output

The demo uses OpenCV window to display the resulting video frame and detections.
It outputs logs to the terminal.

## See also
* [Using Inference Engine Demos](../../README.md)