Doorify

• 

  Notification when a person is detected.

Inspiration

For individuals with hearing disabilities, daily activities that heavily rely on their hearing pose to be a great challenge. While there are current solutions such as visual alert doorbells, they come with limitations. What if the individual is not in the same room as the device? I was inspired by my grandmother, whose hearing has been deteriorating due to her age and rely on mobile calls by delivery personnel for parcel delivery tracking. So I wondered, is there a way we can leverage the wide usage of mobile devices today to solve this challenge?

What it does

This uses computer vision that can be incorporated into camera surveillance systems to identify faces at the door. When a face has been identified, a message accompanied by an image would be sent to the individual.

How we built it

Using facial recognition from Open CV, we are able to identify if there is a person in the view. When a person is identified, then using the Twilio API, a Whatsapp message is sent to inform the individual.

Challenges we ran into

We initially used computer vision to detect humans to activate the sending of the notification. However, during testing, we realised that this causes false positive data if anyone walks past the camera (eg. neighbours). We then decided to use facial detection instead as it reduces the number of false positives since it would mean that the notifications would be triggered only when the front view of a person is detected.

Accomplishments that we're proud of

As our first hackathon, we are proud to be able to put up a product that we see potential in. Furthermore, from the tests that we have done, the product is accurate and is a viable solution to the problem.

What we learned

Understanding the problem that we have at hand is important in order to come up with a solution that fits and value adds to the current situation. For example, this can be easily integrated into current camera surveillance systems to send real-time notifications without having to purchase new hardware.

What's next for Doorify

Besides identifying people at the door, we can also use Open CV to detect parcels and notify the user accordingly. We can also incorporate facial recognition to recognise the people at the door, if they are family members or random strangers so that we do not need to send an image.

Built With

• opencv
• python
• twilio