Model was built using various convolution layers with hyper-parameter tuning to ensure high degree of accuracy.

Accuracy achieved: 98.78% Test data size: 10.3 GB Number of classes used: 20 Language implemented: Kannada

How to run the real time testing??

Clone the repository Create a new environment in conda Install the packages mentioned in requirements.txt Run the command: streamlit run app.py

Enjoy!

Samples of signs and detailed documentation will be uploaded soon.