Integrating Skeleton Based Representations for Robust Yoga Pose Classification Using Deep Learning Models

arXiv — cs.CVFriday, December 5, 2025 at 5:00:00 AM
  • A new study has introduced a curated dataset named 'Yoga-16' aimed at improving automated yoga pose classification through deep learning models. The research systematically evaluates three architectures—VGG16, ResNet50, and Xception—using various input modalities, including skeleton-based representations, which have shown to outperform raw image inputs with an accuracy of 96.09%.
  • This development is significant as it addresses the limitations of existing datasets and enhances the reliability of yoga pose classification, potentially reducing the risk of injuries associated with incorrect postures during practice.
  • The integration of skeleton-based representations in yoga pose classification reflects a broader trend in artificial intelligence, where deep learning models are increasingly utilized for various applications, including medical image classification. This shift underscores the growing importance of transfer learning techniques in achieving high accuracy across diverse fields.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps
Continue Readings
Detection of Cyberbullying in GIF using AI
NeutralArtificial Intelligence
A recent study has focused on the detection of cyberbullying in GIFs using artificial intelligence, highlighting the growing concern of this social issue as it evolves with the internet and social media. Researchers collected a dataset of over 4,100 GIFs from Twitter, utilizing the GIPHY API and applying a deep learning model, VGG16, to identify instances of cyberbullying within these visual formats.
Automated Deep Learning Estimation of Anthropometric Measurements for Preparticipation Cardiovascular Screening
PositiveArtificial Intelligence
A new study presents a fully automated deep learning approach to estimate key anthropometric measurements from 2D synthetic human body images, aimed at enhancing preparticipation cardiovascular screening. The method, utilizing a dataset of 100,000 images, achieved sub-centimeter accuracy with the ResNet50 model performing the best, indicating a significant advancement in the field of sports medicine.
An AI-Powered Autonomous Underwater System for Sea Exploration and Scientific Research
PositiveArtificial Intelligence
An innovative AI-powered Autonomous Underwater Vehicle (AUV) system has been developed to enhance sea exploration and scientific research, addressing challenges such as extreme conditions and limited visibility. The system utilizes advanced technologies including YOLOv12 Nano for real-time object detection and a Large Language Model (GPT-4o Mini) for generating structured reports on underwater findings.