Anthropometry measurements of farm workers using computer vision-based multiview stereo-image sensing

Area of Science:

• Ergonomics and Human Factors
• Computer Vision Applications
• Agricultural Engineering

Background:

• Precise anthropometric data are crucial for designing safe and efficient farm machinery.
• Manual anthropometric measurements are time-consuming, labor-intensive, and prone to errors.
• Developing non-contact measurement techniques is essential for modern ergonomic assessments.

Purpose of the Study:

• To develop and validate a computer vision (CV) based non-contact system for anthropometric measurements.
• To assess the accuracy and reliability of CV-based measurements for stature, vertical reach, trochanteric height, and chest circumference.
• To provide a scalable and efficient alternative to manual anthropometry in agricultural contexts.

Main Methods:

• Utilized an Intel RealSense D435i stereo camera with OpenCV and mediapipe for image capture from multiple angles.
• Recruited 32 participants (16 male, 16 female) for measurements, comparing CV data against manual anthropometry.
• Assessed accuracy using Mean Absolute Difference (MAD) and Mean Absolute Percentage Error (MAPE), and reliability via Intraclass Correlation Coefficient (ICC).

Main Results:

• The optimal measurement distance was 3.0m with a front-facing camera view.
• CV system showed slight underestimation of stature (MAPE 3-4%) with excellent reliability (ICC > 0.90).
• Vertical reach measurements had the largest bias (MAPE 4-5%), while trochanteric height and chest circumference showed minimal bias (MAPE ≤ 4%) with good to excellent reliability.

Conclusions:

• The developed CV system offers a precise, reliable, and non-contact method for anthropometric data collection.
• This technology is scalable and suitable for ergonomic evaluations and improving man-machine compatibility in agriculture.
• The CV-based approach presents a viable and efficient alternative to traditional manual anthropometric measurements.