Sign language recognition

Scope

Detect people’s hands gesture to implement sign language analysis.

Description

First approach:

Human pose estimation: Detects key-points on the body and hands using pose estimation model.
Simple hand gestures related to sign language can be detected.

Second approach:

Background subtraction and contour analysis: train over “Kaggle Hand Gesture Recognition Database”.

Future work

The presence of an expert related with sign languages or the topic is going to be developed is imperative in this kind of projects.
The next step would be to analyze temporal correlation and body correlation to detect a wide range of gestures and sign language.