Entropy-Aware Activity Recognition for Sports: Case Study on Cricket

In the digitalization era, activity recognition substantially impacts everyone’s day-to-day lives. Multiple people and their activities can be seen in the video and their routines depend on video devices, specifically in sports. Video datasets in sports face several challenges, such as intra-class similarity, inter-class ambiguity, noise and motion transitions. The research proposed an entropy-aware activity recognition deep learning method that integrates pre-trained CNN with Density-Based Spatial Clustering of Applications with Noise (DBSCAN) for keyframe selection for Cricket. A synthesized dataset of 25,000 annotated cricket clips is generated across five fine-grained activity classes. These keyframes are uniformly sampled, encoded into CNN embeddings, and clustered using DBSCAN. The chosen keyframes are then fine-tuned via CNN to optimize through categorical cross-entropy loss. The extensive experiments show that the proposed method achieves an accuracy of 96.4% and shows a reliable enhancement of 6% as compared to unclustered data. The results indicate that DBSCAN-driven frame selection enhances generalization across diverse broadcast conditions and provides a scalable explanation for automated cricket video analysis with real-time applications during highlight generation, smart sports surveillance, and players’ performance monitoring.