ASL recognition

Inspiration

Empowering the deaf and dumb, giving voice to their gestures.

What it does

We recognize some standard ASL signs and gestures and convert them into text and voice.

How we built it

We took data from leap motion controller and extracted features for that for a Machine Learning classifier. We then build the dataset for training, and then used a python script to recognize the sign in real-time.

Challenges we ran into

Deciding which features to extract, building those feature, and making our own training data set was challenging.

Accomplishments that we're proud of

Building a dataset with more than 60,000 instances was an accomplishment. This was our first project using Machine learning and a hardware

What we learned

Learned about applying multi-class classification by making test data, and training.

What's next for ASL recognition

Adding more complex gestures which make use of dynamic motion properties of palm.

Built With

Submitted to

HackWITus2
- Winner Amazon Web Services - Best Use of AWS

Created by

I worked on getting data from Leap motion device using Python. Cleaned the features and selected appropriate features for classifying the signs from American sign Language.

Karan Tyagi
I worked on getting and designing the data set from Leap motion controlller using python by incorporating sign language.

HITESH VERMA
Machine Learning Enthusiast | Computer Science Graduate student at Northeastern University
I worked on the python backend processing of leap motion input to construct json and data required for training our network.

Harshad Reddy Nalla
rahulbhat31
Ankur Bambharoliya