Ninja Viz

• 

  Ninja Viz

Inspiration

When we saw the topic was social good, we thought about the visually impaired people, who are having trouble navigating in public places or even in their own home. The visually impaired are more vulnerable to physical injuries, namely accidents. They also have trouble recognizing objects in front of them, which makes them easy to trip over and fall.

What it does

We designed a device that help visually impaired people "hear" the world around them. This can make it possible for them to be more careful when navigating around the house, neighborhood, and any public place. It tells the person what object is in front of them, what direction it is in (to the left, right, or middle), and how far it is away from them (in feet).

How we built it

The first step was to train an object detection model to detect common household objects which took ~8 hours. The model was then optimized to run on the Raspberry Pi and the Edge TPU AI accelerator. The AI model takes in the camera feed and processes it to detect objects and see where they are located. An ultrasonic sensor was then used to measure the distance between the device and the objects. Next, a speaker was connected so that the visually impaired could hear what objects the device was detecting and the distance. Finally a case was 3D designed and printed to hold everything together neatly and compactly.

Challenges we ran into

When we synced the AI model to the audio, the model started to freeze.

Accomplishments that we're proud of

Making a custom computer vision model. 3D printing a custom case.

What we learned

We learned and developed new skills such as working with AI and 3D modeling.

What's next for Ninja Viz

We are hoping to improve this gadget by making a feature where the person can store their friends or family, and it will tell the name of the person (by facial recognition), tell them what direction they are standing, and how far away they are. By also adding text reading capabilities so that visually impaired people could read text not in braille. Adding on to safety, real time map navigation can make it easier and safe to navigate in public places or transportation as well as allowing saved locations to help people navigate around known areas.

Built With

• ai
• css
• html
• javascript
• python