Automated No Ball Detection for Cricket Umpiring: A Vision-Based Solution

This paper proposes a method for detecting no ball, which is an important aspect of every cricket match. The poor decision-making of the umpires in cricket about no ball has been significantly increasing over the past few years. This blunder cost the squad a spot in the tournament's playoffs. The prudent use of technology could prevent controversial umpiring. This proposed model evaluates the optimum strategy by analysing multiple classifiers, largely based on SIFT and ORB by extraction of features. The testing of accuracy, precision, recall, and F1-score of several classifiers such as decision Tree, random forest, SVM, and logistic regression are compared on different classifiers. A decision tree’s accuracy was 90.39%, an SVM’s accuracy was 91.8%, and a logistic regression’s accuracy was 91.5%. According to the outcomes of this research, SVM produces the highest results with 91.8% testing accuracy with SIFT extraction.