iOS Video Recognizer for Blind

Tap and hold your finger on screen to hear what your iPhone see. Use PyTorch's AI to see with your phone. It's demonstrates how to use PyTorch's pre-trained video classification model on a device.

Comment

Inspiration

I like to bring AI tools to disadvantaged people. This spark the light when I see video classification capabilities of PyTorch.

What it does

It uses pre-trained PyTorch video classification model and speak out loud what it see on iPhones.

How we built it

I built it easily using PyTorch iOS sample projects on Github and add it text-to-speech capability of iPhone. When person touch and hold finger on screen, it starts to speak to prevent unnecessary sound.

Challenges we ran into

I need to understand how sample projects work, thanks to PyTorch's developers they are very easy to understand and have a clean code.

Accomplishments that we're proud of

I created this project under two hours.

What we learned

I learned that PyTorch offers great capabilities for on device machine learning.

What's next for Video Recognizer for Blind

I can add image classification model to recognize objects in images and guide user by informing where are the objects.

Share this project:

Updates