PeVL: Pose-Enhanced Vision-Language Model for Fine-Grained Human Action Recognition | Read Paper on Bytez