Unique Approach to Detect Bowling Grips Using Fuzzy Logic Contrast Enhancement

Abstract

Nowadays Cricket has become a much more competitive sport. We can see new bowlers are evolving with their unique bowling styles and variations. A bowler possesses the expertise to bowl multiple categories of bowling in a particular over and baffling the batsman completely. Despite unique bowling styles create confusion in batsmen, the grip of bowlers can reveal greatly what the bowler is trying to bowl. This research concentrates on predicting the type of delivery the bowler is trying to ball with a unique combination of Fuzzy Logic and state-of-the-art machine learning and deep learning models. For the research purpose, a grip dataset is used that contains 5573 images of grips of 13 categories of deliveries. An approach of image contrast enhancement is shown using Fuzzy logic based on the L-channel of the CIE 1976 L*a*b* color space (CIELAB) color space [L*a*b where L=Luminosity and a*b are green, red blue and yellow color] generated from RGB and then the proposed shallow Convolution Neural Network (CNN), VGG 16, KNN, Naïve Bayes, and Decision Tree were trained and the accuracies shown were remarkable.