Seeing Eye

Seeing Eye: ID Objects Visually 📷

Comment

Inspiration

I was inspired to create this website because of an app I heard of that allows blind and seeing people to connect. The app allows the seeing person to help identify what the object they're holding is. The app always had such a large queue of seeing people wanting to help the blind, so I never got a chance to help. However, now I can help way more than one person!

What it does

Seeing Eyes helps the visually impaired identify objects via a webcam. If a person holds up an object to the webcam, the website will identify it via a text blurb at the bottom. Seeing Eyes can recognize plastic bottles, notebooks, pens, scissors, vitamin bottles, gum, cellphones, people, and cups.

How we built it

I used Google's Machine Training technology to train a machine to identify about ten objects. I took about a hundred photos of each object and trained the model. I implemented the ML into a website that I built using HTML, CSS, and javaScript on the Repl.it platform.

Challenges we ran into

I had so much trouble centering the webcam video. The code for the webcam was in the script.js folder, and I tried everything to format it. However, when I did, everything else on the web page was thrown out of wack. So eventually, I just focused on other parts of the website.

Accomplishments that we're proud of

I am very proud of myself for using machine learning on a website. This is only my second time working with machine language, and the first time it actually worked. I am also proud of myself for staying motivated and actually submitting a hack because last year I did not think I had the skills to build anything. I'm just proud of myself for trying and sticking with it. :)

What we learned

I learned that if you put your mind to something and believe in yourself, you can get it done. I also learned that technology is really amazing, and it can be used for a lot of purposes involving helping people.

What's next for Seeing Eye

Next, I hope to implement a voice-to-text feature. The visually impaired might have a hard time reading the screen, so I want the website to read everything, including the item it has recognized, out loud. I had hoped to accomplish that this weekend, but I was unsuccessful. I also want the website to be able to recognize more items, but I was in my college dorm so my selection of items to test was a bit limited.

Share this project:

Updates