Sports-Net18: Various Sports Classification using Transfer Learning

With the advancement of Computer Vision (CV) classification of sports images is very popular nowadays. Deep learning techniques have become feasible to use and see what happens inside the contents of an image. As videos are the collection of continuous images, or `frames', this technique can be also used in videos. Computer Vision tends to reduce the complexities in the human visualization system and understanding by applying convolutional neural network models to accurately recognize and classify objects from the potential and asymmetrical physical world. Classification of images is now very popular in recent years as the emergence of computer vision technology but it is quite challenging. In this paper, we proposed a VGG16 transfer learning model to classify eighteen categories of various sports and we have created our sports dataset which contains 9000 images. Our proposed model has shown a promising result which is 93%.