ASL-T is a data science project that would be used to help translate American Sign Language into the alphabet to make it easier to communicate. The input of this project would be an image and based on the input image, the output would be one of the twenty six letters or maybe nothing if the image is invalid(Doesn’t have a valid ASL sign.)
We will not be hosting the website because our Convolutional Neural Network file exceeds Heroku's storage. But I promise it runs :)
As for the dataset that we will be using to train our model is taken from Kaggle’s ASL dataset.
- Use git LFS to help host our model
- Make our model more robust by finding more data to train it on