iForgot

Inspiration

We can be very forgetful people sometimes. It is always a pain having to manually look all over the house for a lost item. That was when we realized that the power of AI can be utilized to recognize lost items

What it does

This technology can utilize your home security cameras and scan the rooms available to it and highlight your lost item so you know exactly where it is.

How we built it

We used datasets of everyday household items and trained/tested them on our own local machines in order to guarantee their quick and accurate execution on low power machines. No connecting your home security to the internet!

Challenges we ran into

In the time constraint of a hackathon, our machines could not train a single model to recognize all commonly lost household items. We overcame this by modularizing our approach and training small models for a select few items that you may misplace.

Accomplishments that we're proud of

Because of the nature of finding an items from a camera, it isn't especially close to the camera. One technique that came into play fixing this was the sliding window. A small frame inside the image moves across the large frame and looks for the item dynamically in the small window. This makes recognition more accurate from further distances allowing you to find your items even faster

What we learned

We learned a lot about the process of training a local AI model for startup projects. These are invaluable skills in this day and age and will certainly be of use to all of us going forward.

What's next for iForgot

We will be expanding our selection of items that can be searched for and with even more degrees of accuracy. We also want to be able to support taking pictures of your specific items to make the AI able to recognize them even easier.