Overview / Usage

Handling large, heavy, or awkward objects during manual lifting poses a significant risk of musculoskeletal disorders (MSDs) for employees. These disorders encompass various injuries such as sprains, strains, fractures, and soft-tissue damage to the back and shoulders. Surprisingly, more than half of MSD cases result from a lack of awareness regarding proper postures during movement or lifting. To address these occupational health and safety (OHS) concerns, the proposed project aims to employ advanced computer vision and image processing techniques. The objective is to develop a solution that accurately tracks employees’ postures when moving or lifting substantial objects by hand. This innovative system serves a dual purpose: during the training stage, it can educate staff on safe postures for these activities, and in day-to-day operations, it continuously monitors the safety of all relevant employees in their working environments. The project can operate with a standard webcam as the input source. As the solution evolves, there is potential for expansion to incorporate high-resolution cameras, coupled with the processing capabilities of a high-end PC for enhanced performance.

Methodology / Approach

The proposed solution involves the development of a Computer Vision-Based Posture Recognition System. This system leverages advanced computer vision techniques using Python along with MediaPipe for accurate posture recognition. The solution will initially utilize a laptop or computer webcam for real-time monitoring during manual lifting activities.

The system will employ MediaPipe to detect key skeletal points, allowing precise tracking of the user's posture. Python will serve as the primary programming language, ensuring flexibility and ease of integration. The solution aims to provide not only real-time monitoring but also instant feedback to users regarding their posture, promoting corrective actions during manual lifting.

Technologies Used

Technologies:

- Computer Vision
- Machine Learning
- Artificial Intelligence
- Deep Learning
- Wearable Technology
- Web Technologies (HTML, CSS, JavaScript)

Libraries:

- MediaPipe
- OpenCV
- Python Libraries for Machine Learning

Tools:

- Pose Estimation Tools
- Image Processing Algorithms

Software:

- Python Programming Language
- Web Development Tools (for the user interface)

Hardware:

- Standard Webcam (for initial implementation)
- Potential scalability to high-resolution cameras and high-end PCs

Intel Technologies:

- Intel CPU
- Intel Media SDK 
- Intel Python

Documents and Presentations