PoseNet based Model for Estimation of Karate Poses

In the domain of computer vision, human pose estimation is becoming increasingly significant. It's one most compelling areas research, and it's gaining a lot interest due to its usefulness flexibility in variety fields, including healthcare, gaming, augmented reality, virtual trainings sports. Human has opened door opportunities. This paper proposes model for classification karate poses which can be used posture correction trainings. A pretrained model, PoseNet been using results angles between specific joints are calculated fed into K-Nearest Neighbors Classifier classify poses. The obtained show that achieves an accuracy 98.75%.