Leveraging CNN-RF Models for Enhanced Serve Shot Strategies in Table Tennis

The definition of which serve strokes to use depends on the technique practiced and the specific tactical conditions during the game because of the strategic complexity inherent in serve play. This research will use a precise CNN-RF model to identify and classify different server shot motions like backhands, forehands, and volleys. The model categorizes the shots into five classes: topspin, backspin, sidespin, fade spin, and knives spin. The knives pin does not effectively cause the ball to deviate in the air. Reviewing that dataset, including server shots played by pro players, eventually revealed an 87.83% accuracy rate, a testament to how well the model can follow the serve shot motions very accurately. A remarkable accuracy was demonstrated by a model, as evidenced by its 92.81% maximum score for individual serve motions. It was superior in memory, with the correct replay of around 96.23% for the most identifiable serves. The key indicator is the F1-Score, which measures precision-recall balance and is maintained at a high-level during diagnostics of all classes. In macro, the average F1-Score of 87.63 indicates equitable accuracy and impartiality regarding the precision-binary class trade-off. The model was assessed using a dataset of 2670 images, and the results indicated the strong possibilities of its adoption in real-life sports forecasting. The investigation demonstrates that the CNN-RF model performs better than other single CNN or RF models, with observable improvements in accuracy and recall rates. The weighted and macro-averages ended the last argument related to the potential effectiveness of the proposed model in handling different sets of classes that are biased toward some classes.