Overview / Usage

This demo can be applied in different places. In school, every student has their pictures in the system, and this can be used as ID identifier and Access control to reduce waiting lines and increase security level. For smart home, you can register your family members in the system with pictures, and then if the camera detect stranger beside your door, it will send a text message directly to the user. In factories, this can be used as intruder detection, where supervisor will be alert when someone enters the restricted area.

Methodology / Approach

The original code is for smart classroom, and I implement some functions based on the original demo. This demo used openvino pretrained models based on MobileNet V2 backbone and RMNet backbone. It can detect faces and tell if the person is in the system by using face re-identification model.

I used libcurl to connect to one SMSGateway, and send the text message to user's phone immediately as an alert when necessary.

Still working on creating user interface and adding two-way message system to make it more interactive.

Technologies Used

Software:

Pre-trained face reidentification model

Pre-trained face landmarks model

Pre-trained person-action detection model

Pre-trained person face detection model

OpenVINO Tool Kit (Model Optimizer)

Libcurl

​

Hardware:

Camera

Neural Compute Stick 2

Up Squared Box/Board