Normally, when we train any Machine Learning algorithm, we require lots of similar type of data for the algorithm to train. But, this approach is not effective for tasks like face recognition. Face Recognition aims not only to detect a human face in a given image, but also to recognize whose face it is in the image. But to train such an algorithm in the traditional approach, we will require tens, if not hundreds of images of each person as a different class and train the algorithm to learn to classify the faces. This is not practical as one might not be able to get so many images of a single person. In this article, we imply One Shot Approach for image recognition. This approach enables us to recognize faces using a single image for each person.
Continue Reading