Authors:
K. N. V. Satyanarayana, Yarlanki Kushanth, Lingam Sai Geethika Thanmayie, Gudipudi Sai Babu, Kodiboyina Ramya Sri, Maruf Farhan
Addresses:
Department of Electronics and Communication Engineering, Sagi Ramakrishnam Raju Engineering College, Bhimavaram, Andhra Pradesh, India. Department of STEM, University of Sussex International Study Centre, Brighton, England, United Kingdom.
Abstract:
People exercise to keep healthy. Many of them work out without a trainer, which can lead to poor posture, muscle injuries, and long-lasting injuries. People who work out typically only watch videos online. Do what they think is right; no one tells them when they are doing something wrong. To combat the problem, researchers design a system that observes how people work out and assists them in correcting their posture in real time. This technology records camera footage and develops algorithms to detect key body parts, such as shoulders, elbows, hips, and knees. It uses these points to model the person's body and calculate joint angles. Then it compares those angles to the posture guidelines for each exercise. A notification appears on the screen if their posture is incorrect, allowing them to change it immediately. Researchers designed the system using recordings of people exercising with poor posture to learn how people move and how the system will behave in various situations. Researchers found that the method worked effectively with simple activities in low light and a crowded background. It worked well on a normal computer and was accurate 98%–99% of the time. It is inexpensive and easy to use because it only requires a camera. It helps home, gym, and sports athletes train more effectively and avoid injuries. Safe exercise will help injured persons heal.
Keywords: Deep Learning; Pose Detection; Computer Vision; Real-Time Monitoring; Posture Correction; Exercise Tracking; Human Activity Recognition.
Received on: 15/12/2024, Revised on: 06/02/2025, Accepted on: 01/05/2025, Published on: 03/01/2026
DOI: 10.64091/ATICL.2026.000254
AVE Trends in Intelligent Computer Letters, 2026 Vol. 2 No. 1 , Pages: 42–52