Overview / Usage

This model is a simple model for personal use where digits can be recognised form hand gesture.

This model can be extended for other applications like recognising English alphabets from hand gesture with the appropriate dataset and minor changes.

Methodology / Approach

First, I added some of my images using a webcam and preprocess them to the desired dimensions and size and stored them in the appropriate folder. Then trained a CNN on dataset+my captured images for better accuracy.

The Model Architecture: Conv->MaxPool->Conv->MaxPool->Flatten->Dense(512)->Dense(128)->Output(10)

Technologies Used

API-Keras

Backend-Tensorflow

OpenCv and other python libraries.

Python>=3.6

Repository

https://github.com/sakshamj74/Digit-Recognizer-From-Hand-Gesture