Action Recognition Algorithm Exploration Based on Deep Learning
Yaguang Liu · 2024
Basketball technology's action recognition algorithm is crucial, though it encounters inaccuracies in performance positioning. The conventional ant colony algorithm doesn't resolve these recognition challenges effectively. Consequently, a genetic algorithm tailored to this domain is proposed for enhancement. Initially, the theory of biological evolution identifies influential factors, segmenting indicators to minimize disruptive influences in the action recognition process. Subsequently, a genetic algorithm-driven solution for action recognition is devised and its outcomes are rigorously analyzed. MATLAB simulations demonstrate that, under specific evaluation criteria, the genetic algorithm surpasses the standard ant colony approach in accuracy and processing time for action recognition variables.