Automated Parental Control

unsuitable content notifications
Theme
Devfest2021

💡 Inspiration

Seeing kids grow around smart home appliances, we notice often how much harmful content can be at their disposal, harm that is not only physical but also extends to their mental well-being. physical harm which affects their vision and is the cause No1 that leads to sight deterioration among children, in addition, kids are exposed often to inappropriate content on TV that involves language, horror and other unsuitable shows.

🔍 What it does

our solution, & through a camera integrated in the home TV allows to recognize the child and it's inputted age who is watching & then notifies the parent in case the content he is viewing is unfitting, In addition if the camera detects that the kid is at a close distance from the TV screen that might hurt his sight, it blurs out the screen & displays through animations that he is too close & should stand back to clear the screen again.

⚙️ How we built it

we used computer vision services provided by azure cognitive services to deploy an AI model that allows to do facial recognition & associate it with it's input age, and then through the metadata of the current show; decides to whether notify the parent through an app of the unsuitable content the child is watching or not. Computer vision also helps to calculate the distance from which the child is to the TV screen and blurs the screen as a preventive measure. The app can be extended to analyze if the child shows symptoms of eyesight deterioration through analysing eye movements.

🚧 Challenges we ran into

Among challenges we ran into, is that the LG API's are only available to Business contracts, so not having access to such useful ressources is a little bit of a barrier in our way but we opted for simulating the services offered by LG by other available APIs like azure cognitive services & python libraries like OPENCV

✔️ Accomplishments that we're proud of

Among the accomplishments we're proud of are :

Implementing a functional model of distance-to-screen detection that blurs out the screen when a viewer is too close within a defined range.
- Implementing a functional model that detects emotions of a viewer and sends notifications to the monitoring app,like if the spectator is exposed to fearful or harmful content.

📚 What we learned
- using azure services & integrating them using python.
- using the opencv library in python.
- an overview of the available APIs LG offers in it's smart home appliances ecosystem