A novel Artificial Intelligence approach for classification of footprints based on the size has been proposed. Our main objective is to develop a system for person identification similar to the systems developed for fingerprints. Making a system that checks the input image with each and every image in the database would have been a very computationally expensive task. So, we used an Artificial Intelligence technique called Convolutional Neural Networks(CNN), in order to classify the image into two class i.e. Small and Large, three class namely, Small, Medium and Large and seven class classification as in US size 4, 5, 6, 7, 8, 9, 10. A paper scanner was used to obtain the images to classify the footprints. 880 footprint images from 88 people (5 samples per foot) were analysed, leading to a conclusion that footprints can be classified according to shapes using Convolutional Neural Networks architecture of Deep Learning. Footprints could be classified into 3 main features: Shape or Geometry Features, Eigen Features and Minutiae Features. We have worked on Shape Features of foot which is used for classification purpose only. We can reduce sample size by using shape features. Then we can pass the classified image for the extraction of minutiae which will take less time to process the query image.