DeepASL uses a webcam video feed and Python to interpret American Sign Language (ASL) hand gestures in real time. The goal of this project was to learn the fundamentals of Convolutional Neural Networks and to understand how Computer Vision can make them interactively useful in real-world applications.
-
Clone the repository:
git clone https://github.com/cesarealmendarez/DeepASL.git -
Navigate to the project directory:
cd DeepASL -
Install the required packages:
pip3 install opencv-python mediapipe numpy -
Run DeepASL:
python3 app.py
Once you run DeepASL, two windows will appear:
- Analytics Window: Displays the raw video feed along with extracted data points used to interpret hand landmarks, steadiness, depth perception, output confidence, and snapshot triggering.
- Hand Segmentation Window: Shows how the network breaks the image down into a pattern of 1s and 0s, prompting its best attempt to guess which ASL letter you are showing.
- MediaPipe: Used to detect the shape of the hand and create a skeleton-like outline for segmenting useful classification features.
- MNIST Handwritten Digits Classification using a Convolutional Neural Network (CNN)
- A Comprehensive Guide to Convolutional Neural Networks — the ELI5 Way
- Simple Introduction to Convolutional Neural Networks