Sign Language and Gesture Recognition for Disable Persons

These is the GUI made for User Interface user don't need to go for code one can by clicking button open software. GUI is still in Progress
These is the picture showing where we are collecting the key-points/hand landmarks for converting it into an Array and use it
These is our Data Set which we collected manually by clicking the photos

Inspiration

Smart India Hackathon was a platform where we got such interesting topics. We see these people face many problems but we thought about how our DeepLearning knowledge can fix that

What it does

It does is it can recognize the Indian Sign Languages and has the ability to convert it into a speech. We are still working on the speech for accuracy.

How we built it

We first reached for Deep Learning and saw how we could train perfect data, then with the help of Python libraries like Keras, Media Pie, and Cvzone we started training the data and collected huge data sets. We trained a huge data set and built a model called "Bhasha.h5".

Challenges we ran into

We faced many challenges training a huge data isn't simple accuracy was fluctuating, the biggest challenge for the Text-Speech was how to take all words properly and integrate them to create a proper Word. So we defined a time function logic and tried to implement it in the Program. Still, there was a lack of accuracy in building a word so we drove the idea of connecting these programs with ChatGPT 3.5 with an API key for accuracy.

Accomplishments that we're proud of

We accomplished in a very minimum time such a big program with a well-trained big data set

What we learned

We learned how to not fear encountering and facing problems coming towards us. We also learned that everything can have a solution just we have to think calmly and work hard.