ASL-Translator

Inspiration

Inspiration for our project was to find a user-friendly and intuitive way for people to comprehend and translate American Sign Language to English words

What it does

The web-application uses Machine Learning techniques to detect and classify American Sign Language to English Alphabet with 90% accuracy.

How we built it

In the Google Cloud Platform, we uploaded a dataset of different signs in ASL corresponding to each letter in the alphabet. We created this by using Google AutoML Vision, which allowed to create a customized Machine Learning model that we had to train, evaluate, and test, in order to achieve a satisfactory accuracy of predictions.

Challenges we ran into

We were not able to make our ML model to showcase the predictions in the front-end. We tried to learn how to link the back-end technologies with the front-end, but were unsuccessful in doing so.

Accomplishments that we're proud of

The model correctly predicted the ASL alphabet letter with an average accuracy of 90%.

What we learned

Leveraging Google Cloud Platform to build custom Machine Learning Models.

What's next for ASL-Translator

Real-time motion-detection for ASL alphabets, allowing easy conversation between American sign language users and non-users of ASL.

Built With

• cloudvisionapi
• css3
• express.js
• googleautomlvision
• googlecloudplatform
• html5
• react