The most important concern for the parents is their children. This concern takes the form of several questions in the mind of parents - Are their children coming back home at correct time?, Are they hanging out with their friends after school? or Are they leaving the home without letting their parents know? However, it is impossible for the parents to always keep an eye on their children for several reasons:
1) Too busy in work that they can't arrive home earlier or at the same time with their children
2) School bus service, parents don't pick up their children by their own
3) The teenage issue is common, teenagers seems hate to be tracked by parents.
Fortunately, it is possible to solve the above problem by using mobile technology and virtual computer. Our project provides a solution for parents to keep an eye on their teenagers, the only thing you need is a webcam and a smartphone connected to the Internet.
The goal of our Adolescent Monitoring System is to track the home arriving time of teenagers and keep parents informed when their children come back home or leave home to hang out with their friends. Figure 1 shows the prototype of our project, including a virtual door and email notification system.
In the Internet of Things, sensors are designed to communicate with each other via the Internet. In our project, users first provide several pictures of their teenagers for the system as the face database. The virtual door has been programmed to sense when someone come in/go out and to trigger the web camera inside the peephole to take a picture of the person. Then the face verification part will inform parents that who has already come back home or who is leaving to hang out with friends.
Based on the aforementioned approach, we built the prototype of Adolescent Monitoring System to illustrate our idea.
We used Grove Starter kits, Intel Edison board, a USB webcam and a smartphone to implement the prototype. We also made a virtual door to simulate the real world situation. Two rotary angle sensors work as door handles, the webcam is installed in the peephole to snapshot, the smartphone is used to receive messages and email with snapshot attached (Figure 2 shows the installation). When the family member comes from outside and rotates the handle of the door to unlock it, the camera will take a snapshot of him/her. Then our system will do the face verification using the images in DataBase. If it finds any matching image, it will send an email with a message "(family member's name) coming in". If no image matches, the parents will receive a message "someone coming in" and then they can check who has entered in the mailbox. If someone is going outside of the home, the parents will receive a message "someone going out", snapshot attached in the email.
Our project is distributed in 6 steps:
- configuring Intel Edison board
- mounting the base shield on Edison board
- connect rotary angle sensor to base shield
- connecting webcam to Edison board
- running the python file on Edison board
- putting everything inside a door-like box
In the beginning, the video provides the demo on how the system works. In the database, there are several images of our team members (Changxin Bai, Lanyu Xu and Mohit Kumar). Our testing result is shown in Figure 3.
Our main difficulty here is the face verification part. Current algorithms for face verification still need more training and test, so do ours. During our training process, we find the result can be influenced by many variables, like contrast, facial expressions, etc. We need to set the threshold around 55% to get the matching item and ensure the accuracy. In the future, we will focus on the face traning part, try to use multiple pictures of the same person to improve its performance.