Intel® Edge AI Scholarship Program - Project Showcase

Project Showcase: Mask Detection in Public

Preface:

In early 2020, Coronavirus disease (COVID-19) outbreak. The virus is highly contagious through respiratory droplets produced when an infected person coughs or sneezes. For the reason, many Asia country's governments has regulated people to wear masks in public places. In the situation, the mask detection app based on AI edge computing device might be a useful tool for supporting the policy.

Ideal Application:

Through AI OpenVINO toolkit, develop an effective model to detect whether people wear masks correctly or not in the public.

Ideal Example Demo (Based on the Mask Detection results, show different colors of square):

(※ Yellow: wear in corret way, Light Green: not wear a mask, pink: wear incorrectly)

Original Open Resources: https://www.facebook.com/groups/525579498272187/permalink/642268049936664/

Actual Development Result

Development Details

Model used: face-detection-retail-0005

Data used: *Open Source Dataset (00e992ca.jpg)

Development tool: Google Colab

This Project Usage

Changeable Codes:

1. Downloading Model Part

You may change the argument, name, to the model intended to download

!python $model_zoo'tools/downloader/'downloader.py --name face-detection-retail-0005 --precisions FP32-INT8 -o /content/OpenDevLibrary/demo_files/models

2. Running Inference Part

You may change the arguments, i for input_image, t for detect_purpose, m for model usage.

!source /opt/intel/openvino/bin/setupvars.sh && python app.py -i "/content/OpenDevLibrary/demo_files/images/00e992ca.jpg" -t "FACE" -m "/content/OpenDevLibrary/demo_files/models/intel/face-detection-retail-0005/FP32-INT8/face-detection-retail-0005.xml"

Project To-Dos

1. Apply orignial model to transfer learning and fine tuning on "Mask Detection Dataset" (open-source dataset)
2. Adopt softmax activation function for detecting different situations of wearing masks
3. Deploy on edge devices to realize edge computing

REFERENCES:

OpenVINO Edge AI Applications deployment on Google Colaboratory

https://github.com/alihussainia/OpenDevLibrary

Intel's Official Installation Guide for OpenVINO on linux:

https://docs.openvinotoolkit.org/latest/_docs_install_guides_installing_openvino_linux.html#install-openvino

Model used in Demo.ipynb i.e. face-detection-retail-0005

https://docs.openvinotoolkit.org/latest/_models_intel_face_detection_retail_0005_description_face_detection_retail_0005.html

*Open Source Dataset

https://www.facebook.com/groups/525579498272187/permalink/642268049936664/

Projects Contributor

Po-Chuan Chen, Li-Cheng Hsueh