Basketball technical action recognition based on a combination of capsule neural network and augmented red panda optimizer

Basketball is a group sport that needs precise identification of the players’ practical actions in different shooting movements for effective training and performance enhancement. This subjective nature of training assessments that most of the time rely only on coaches’ observations, highlights the need for objective analysis tools. The subjective and non-objective nature of present educational calculations that are often based on the observations and experiences of coaches and coaches, highlights the requirement for objective and data-driven analysis tools. Such tools can help trainers make more precise and unbiased calculations of student performance and make better instructional choices. This study presents a new model to identify the basketball technical actions based on combination of the CapsNets or Capsule Neural Networks with an ARPO or augmented variant of Red Panda Optimizer. The study conducts the tasks presented by changing lighting settings and complicated human movements in basketball. By means of the suggested CapsNets/ARPO model, the network’s capability can be improved in distinguishing the dynamic targets. The CapsNet/ARPO system reaches advanced performance in the recognition of shooting actions in basketball, with an accuracy of 92.6% and outperforming existing approaches. Its modular design and user-friendly interface make it easily integrable, and a case study with a professional team indicates significant improvements in player performance (15.6% increase in shooting accuracy) and reduced implementation time (30%) to demonstrate its potential to improve basketball analytics and coaching.