SAFELIFT: Safety-Aware Feedback for Ergonomic Lifting & Injury-Free Tasks

Abstract

Work-related musculoskeletal disorders, often caused by unsafe lifting techniques, remain a persistent threat to worker health and safety. We present SAFELIFT, a safety-aware recommender system that automatically detects risky lifting behaviors and generates corrective feedback. Using monocular video input, SAFELIFT extracts ergonomic parameters to compute the Lifting Index (LI) from the Revised NIOSH Lifting Equation. When the LI exceeds a safety threshold, the system produces both graphical and textual recommendations to promote safer postural strategies. Unlike prior approaches, SAFELIFT requires no wearable sensors or multi-camera setups, enabling scalable and low-cost deployment in workplace environments. To assess its effectiveness, we conducted a two-phase evaluation: (1) domain experts (ergonomists, occupational safety professionals, medical staff) assessed the accuracy and relevance of the recommendations, and (2) lay users evaluated different presentation formats, judging their clarity, helpfulness, and trustworthiness. By integrating ergonomics with recommender system design, SAFELIFT contributes to a new class of context-aware, safety-oriented recommendation technologies for occupational health.