Innovative Approaches for Action Detection: Leveraging CNN-RF in Cricket Umpire for a Gesture Recognition

In the article, the author gives a deep analysis of the model operation. This is done by employing a lot of performance indicators that allow testing of how reliable and accurate the classification system is. This finding brings a sign of hope, with the accuracy ranging from 78.10% to 90.83% on the different lines of action. Class 3 is above the rest in terms of accuracy with a rate of 85.76%, and with a recall rate of 94.87% and an F1-score of 90.09%, it is also leading in the overall list. On the contrary, Class 4 has a surprising 74.2% recall rate and receives an F1 score of 80.7%. The macro, micro, and weighted values represent the model intermediate performance data (model averages). The mean macro results for precision, recall, and F1score are 84.96%, 85.79%, and 85.13%, respectively. The weighted average accuracy, recall, and precision measures show a slight enhancement, meeting 85.44%, 85.06%, and 84.98%. Therefore, the model showed flexibility in taking care of samples being unequally represented. Very much as often as expected, and regardless of weights, the micro-average accuracy, recall, and F1-score are all 85.06%. This points out the model’s equal observance of acknowledging umpire actions. In addition, the model is very accurate overall, achieving 85.06117% accuracy, which shows its reliability and effectiveness when it comes to sports umpire movement recognition in any environment. This section presents the abstract, which gives a summary of the statistical and practical results of the model. It explains how the model learned to differentiate cricket umpire actions and how it was helpful in improving decision-making when it came to cricket officiating.