Inspiration : Blind or Deaf always not ignored with the cutting-edge technology. We want to bring all technological experience by standard finger language. Also, people working in high noise environment or Biomedical lab , where employees are prefer to communicate remotely by finger language.

What it does : Python media-pipe can detect co-ordinates of the finger language. Pass these co-ordinate to Pre-trained ML model to build machine understandable language and send to IOT or Google Voice.

How we built it: MediaPipeline will capture finger position and co-ordinate. Once co-ordinates have enough data, pass that information to ML model to build command.

Challenges we ran into : Don't have enough system capacity to run ML model with 1000+ epoch. Will try to train model as best as I can

Accomplishments that we're proud of : UI with Hand gesture is captured and position to communicate / browse google.

What we learned : Learn new technology like Tensorflow , MediaPipeline. Handled huge dataset with 1600 columns .

What's next for Finger Language Recognization : Rewrite the model in Tensorflow lite. Integrate with IOT device like "Rasberry PI" / BioMedical Monitoring system / Robotics System

Built With

Share this project:

Updates