Nowadays, Facebook and Google make advertising extremely personalized. Looking at a product meant your Facebook feed being filled with directed advertising. But what if this is brought to a whole new level, where directed advertising is done on the non-digital world.

The ideas were gathered from famous movies such as the Minority Report and the Ghost in the Shell, where displays around the city are interacting with the user.

What it does

iCMyself is a facial recognition software that can be deployed on to digital glass (or currently projected onto semi-transparent glass) with the goals to target information towards the person looking through the glass. This can be in the form of weather information or directed advertising, such as changing a shopfront window dynamically based on the person looking at it.

The message currently showing on the application dynamically generated based on time of the day, the weather outside, and the person viewing the display. This message gives recommendations for how users should prepare themselves for the day.

How we built it

Using the MagicMirror core. We have added the facial recognition modules and weather forecast modules (using The Weather Network api). The code is written in Node.Js and Python.

Challenges we ran into

The open source libraries were never MANAGED!!! We had to re-construct libraries from scratch, causing extra cost on time.

Cross-platform between Mac OS, Windows and Linux was pain (Yes! it works on all three machine!). But Python libraries were all different across all three platforms.

Accomplishments that we're proud of

Facial recognition actually recognize faces. It works on all three platforms (Mac OS, Windows, Ubuntu Linux).

What we learned

DO NOT!!! RELY on open source libraries.

What's next for iCMyself

  • Proper facial recognition using proper visual recognition engine (Ex: Watson)
  • Expand facial recognition database to Facebook
  • More customized information/advertisements
  • Implement other forms of AI such as voice enabled AI assistant to interact with the user
  • Facial expression analysis to indicate user feedback to items on the display (indicates of marketing products being shown)
  • To make it enterprise grade for commercial use

Built With

Share this project: