Occlusion-Aware Worker Detection in Masonry Work: Performance Evaluation of YOLOv8 and SAMURAI DOI Creative Commons
Sukeun Yoon, Hyunsoo Kim

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(7), P. 3991 - 3991

Published: April 4, 2025

This study evaluates the performance of You Only Look Once version 8 (YOLOv8) and a SAM-based unified robust zero-shot visual tracker with motion-aware instance-level memory (SAMURAI) for worker detection in masonry construction environments under varying occlusion conditions. Computer vision-based monitoring systems are widely used construction, but traditional object models struggle occlusion, limiting their effectiveness real-world applications. The research employed structured experimental framework to assess both brick transportation laying tasks across three levels: non-occlusion, partial severe occlusion. Results demonstrate that while YOLOv8 processes frames 2.5 3.5 times faster (28–32 FPS versus 9–12 FPS), SAMURAI maintains significantly higher accuracy, particularly conditions (92.67% 52.67%). YOLOv8’s frame-by-frame processing results substantial degradation as severity increases, whereas SAMURAI’s memory-based tracking mechanism enables persistent identification frames. comparative analysis provides valuable insights selecting appropriate technologies based on specific site requirements. is suitable characterized by minimal occlusions high demand real-time detection, more applicable scenarios frequent require sustained activity. selection an model should be initial assessment environmental factors such layout complexity, density, expected frequency. findings contribute advancement reliable enhancing productivity safety management dynamic settings.

Language: Английский

Performance Analysis of Wearable Robotic Exoskeleton in Construction Tasks: Productivity and Motion Stability Assessment DOI Creative Commons

Ju-Taek Oh,

Gu-Young Cho,

Hyunsoo Kim

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(7), P. 3808 - 3808

Published: March 31, 2025

The construction industry is physically demanding, often requiring workers to lift heavy materials, perform repetitive bending motions, and maintain stability on elevated structures. Wearable robotic exoskeletons have been introduced as a promising solution alleviate physical strain enhance work efficiency. However, prior research has predominantly focused the ergonomic benefits injury prevention potential of exoskeletons, with limited quantitative analysis their impact actual productivity. This study addressed this gap by experimentally evaluating effects wearable exoskeleton productivity motion stability. A total 20 experienced participated in controlled experiments involving three representative tasks: sack carrying, masonry bricklaying, scaffolding installation. Each task was performed under both low-intensity high-intensity conditions, without exoskeleton. Performance metrics, including output, movement stability, postural control, were measured using IMU sensors tracking over 2 h period. results demonstrated that exoskeleton-assisted led significant improvements, particularly tasks, gains up 59.5%. Additionally, metrics showed 24.8% 35.4% reduction sway areas, indicating enhanced balance control. findings further revealed advantage increased time, highlighting mitigating fatigue during prolonged sessions. These provide empirical evidence can serve effective tools for improving worker positioning them viable solutions demanding tasks related industries.

Language: Английский

Citations

0
Occlusion-Aware Worker Detection in Masonry Work: Performance Evaluation of YOLOv8 and SAMURAI DOI Creative Commons
Sukeun Yoon, Hyunsoo Kim

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(7), P. 3991 - 3991

Published: April 4, 2025

This study evaluates the performance of You Only Look Once version 8 (YOLOv8) and a SAM-based unified robust zero-shot visual tracker with motion-aware instance-level memory (SAMURAI) for worker detection in masonry construction environments under varying occlusion conditions. Computer vision-based monitoring systems are widely used construction, but traditional object models struggle occlusion, limiting their effectiveness real-world applications. The research employed structured experimental framework to assess both brick transportation laying tasks across three levels: non-occlusion, partial severe occlusion. Results demonstrate that while YOLOv8 processes frames 2.5 3.5 times faster (28–32 FPS versus 9–12 FPS), SAMURAI maintains significantly higher accuracy, particularly conditions (92.67% 52.67%). YOLOv8’s frame-by-frame processing results substantial degradation as severity increases, whereas SAMURAI’s memory-based tracking mechanism enables persistent identification frames. comparative analysis provides valuable insights selecting appropriate technologies based on specific site requirements. is suitable characterized by minimal occlusions high demand real-time detection, more applicable scenarios frequent require sustained activity. selection an model should be initial assessment environmental factors such layout complexity, density, expected frequency. findings contribute advancement reliable enhancing productivity safety management dynamic settings.

Language: Английский

Citations

0