Hi! Welcome to my first computer vision project.

What it does

As the name suggests this is a simple sign language detector that can be used for people who are learning American Sign Language (ASL).

How it does it

Using python, meadiapipe and opencv I was able to analyse data from the webcam and also identify landmarks on the hand that can be used to create datasets of x and y coordinates. Using these datasets, I was able to train a random forest classifier using scikit-learn to create a sign language model that achieved a 100% cross validation score. Putting it all together, the model can be used in real time in conjuction with opencv to detect hand signs.

How to try it out

1. Pull the project/Download a zip and open in an IDE of your choice
2. Run the files in the order
   1. collect_imgs (Follow the steps on screen to collect images)
   2. create_dataset
   3. train_classifier (This should reveal the classifier score in the console)
   4. test_classifier (Final file that should run the model in real time)

Example Images