Computer vision is an ever-growing and never-ending technology covering a broad spectrum of niche areas such as object detection and recognition, human action recognition, and so on. Since the last decade, there has been an increase in demand for computer vision models which is helping mankind in making human life better, safer, and easier. Human action detection and recognition will be helpful in building healthcare, sports analysis, surveillance, human-computer interaction, and more. The task of estimating the spatial position and orientation of objects or body parts in an image or video is known as pose estimation. Computer vision-based pose estimation models concentrate on determining the pose of human subjects in multimedia data, like images or videos. This can involve the process of estimating the locations of body joints or important body parts. These models are frequently employed in fields like human-computer interaction, gesture recognition, and action recognition, among others. In this paper Mediapipe pose estimation, YOLOv7 and YOLOv8 models were studied and it has been observed that YOLOv8 outperforms by giving mAP of 0.995 for FaceBlockLeft and StomachBlockRight. The proposed model was able to get one recall for both these moves.