The SEWAbility system: a video-based job analysis framework for understanding task-specific job demands.

Area of Science:

• Human-Computer Interaction
• Biomechanical Engineering
• Artificial Intelligence

Background:

• Objective job analysis is crucial for vocational rehabilitation and employment.
• Current methods often lack objective, task-specific biomechanical data.
• AI-powered video analysis offers a novel approach to quantify job demands.

Purpose of the Study:

• To introduce the Smart Evaluation of Work Ability (SEWAbility) system for objective job analysis.
• To demonstrate the derivation of task-specific biomechanical benchmarks from workplace videos.
• To explore the potential of SEWAbility in supporting employment recommendations and vocational rehabilitation.

Main Methods:

• Utilized an AI-powered video analysis system (SEWAbility) processing factory surveillance footage.
• Employed a hierarchical pipeline: video clustering into tasks, subdivision into cycles and elements.
• Extracted 88-dimensional handcrafted features and Repetitive Motion Pattern (RMP) features.

Main Results:

• SEWAbility successfully distinguished between different sewing task types (tops, beddings, bottoms).
• Accurate identification of work cycles and extraction of repetitive work elements were achieved.
• Generated objective, interpretable, task-specific biomechanical benchmarks from motion data.

Conclusions:

• SEWAbility enables scalable and systematic quantification of job demands from workplace videos.
• The system demonstrates feasibility in deriving task-specific biomechanical benchmarks.
• SEWAbility shows promise for data-driven vocational rehabilitation and employment support for individuals with disabilities.