Abstract : This project was created intending to use computer vision to be able to recognize the Sign Language in real time with high accuracy. The reason for such a project is to help diminish the gap between those who can hear well and those hard of hearing or even deaf. This can be overcome by creating a dataset of images that correspond to the alphabets, digits or signs applied to deep neural networks. These images are labeled according to the letter being signed. They are processed through a neural network using transfer learning to help the machine “learn” what is being signed after already having been taught on larger datasets of many more images and classifications.