International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Real-Time Posture Detection for Effective Workouts Prof. Sourabh Natu1, Mohit Kesare2, Dhruv Revar3, Sawarmal Kumawat4 1 Professor, Dept. Of Computer Engineering, TSSM BSCOER, Maharashtra, India 2 Dept. Of Computer Engineering, TSSM BSCOER, Maharashtra, India
Dept. Of Computer Engineering, TSSM BSCOER, Maharashtra, India 4 Dept. Of Computer Engineering, TSSM BSCOER, Maharashtra, India ---------------------------------------------------------------------***--------------------------------------------------------------------3
Abstract - This research presents an innovative approach
enabling users to receive continuous feedback throughout their workout sessions.
to address the challenge of maintaining correct body posture during exercise routines through the utilization of artificial intelligence (AI) and computer vision technologies. Correct posture is essential for optimizing the effectiveness of workouts and reducing the risk of injury, yet many individuals struggle to accurately assess and maintain proper form. The proposed system, employs advanced pose estimation algorithms to detect exercise form at real-time. By capturing video input from the user's computer camera and processing it to detect key body points, it offers immediate insights into the correctness, alignment, and stability of the user's posture. Additionally, the system compares the detected pose with a library of reference postures, enabling users to fine-tune their form for optimal results.
2. LITERATURE REVIEW Sağ et al. [1] (2018) introduced a novel Kinect-based system capable of real-time posture analysis during exercises. By leveraging depth-sensing technology, the system achieved high accuracy in detecting key body points, allowing for precise assessment of exercise form. Jafari et al. [2] (2020) proposed a smartphone-based solution that utilized accelerometer and gyroscope data to assess posture during various activities, demonstrating promising results in terms of accuracy and usability. Liu et al. [3] (2021) explored the use of convolutional neural networks (CNNs) for real-time detection of exercise actions, showcasing the potential of AI in fitness applications. Liang et al. [3] (2019) proposed a reinforcement learning framework for adaptive exercise coaching, where the system learns from user feedback to optimize exercise routines over time. He et al. [4] (2019) employed deep learning techniques to develop a robust system for recognizing yoga poses from video data. Their approach, based on convolutional neural networks (CNNs), achieved impressive accuracy rates and demonstrated the potential for automated pose recognition in fitness applications. Ma et al. [5] (2020) developed a virtual coach system that analyzes user movement patterns in real-time and provides actionable feedback to improve exercise form and performance. Zhang et al. [6] (2020) introduced a novel method for multi-person pose estimation using a single RGB camera. Their approach, based on a combination of convolutional neural networks (CNNs) and geometric constraints, achieved state-of-the-art results in real-time pose estimation, making it suitable for applications such as fitness tracking and augmented reality. Wang et al. [7] (2019) proposed a hierarchical attention-based network for action recognition in fitness videos. By incorporating spatial and temporal attention mechanisms, their model achieved superior performance compared to traditional CNN-based approaches, particularly in scenarios with complex motion patterns and background clutter.
Key Words: Pose Estimation, Posture Detection, Computer Vision, OpenCV.
1.INTRODUCTION In the realm of fitness and exercise, maintaining correct body posture is fundamental to maximizing the benefits of workouts and minimizing the risk of injury. However, for many individuals, achieving and sustaining proper form can be challenging, especially without the guidance of a trained professional. Traditional methods of posture assessment often rely on subjective observations or costly personal training sessions, limiting accessibility and scalability. To address this challenge, the project "Real-Time Posture Detection for Effective Workouts" introduces an innovative solution that leverages real-time posture detection using advanced artificial intelligence (AI) and computer vision technologies. This project aims to provide users with immediate feedback on their exercise form based on the analysis of key body points, without the need for additional equipment or human intervention. By capturing video input from the user's camera in realtime, the system detects key body points and identifies specific exercises or poses being performed. Unlike traditional methods, which may rely on static images or manual assessments, our system operates dynamically,
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 1763