From Periodic Inspection to Continuous Monitoring: A Case Study of Camera-Based Joint-Angle Alerting for Ergonomic Risk on a Printed Circuit Board Assembly Line

Ergonomic risk management in electronics manufacturing still relies largely on manual, periodic inspections that offer little continuous visibility of operators' postural exposure. This case study reports the design, implementation and pilot-line validation of a Web-based ergonomic monitoring platform at an electronics manufacturing plant in Manaus, Brazil, developed over a seven-month applied research and development cycle (January–July 2026) by a technology institute in partnership with the manufacturer. The platform detects operators in industrial camera streams with a YOLOv8l detector, extracts body and hand landmarks with MediaPipe Pose and Hand Landmarker, computes joint angles, classifies each monitored joint into safe, light, moderate or high-risk angle bands with ordinal scores, and raises an alert only when a high-risk angle persists beyond a configurable holding time. Angle bands and persistence rules were defined with an NR-17-oriented ergonomics consultancy and are editable through the interface. Events flow through a containerised micro-service pipeline (RTSP ingestion, RabbitMQ messaging, REST API, SQL Server) to a Web interface with real-time monitoring, dashboards, PDF reports and a 3-D skeleton view. On a 500-frame sample from the surface-mount technology line, compared with human inspection, operator detection achieved 94.5% precision, 98.2% recall and a 96.3% F1-score; the reported hit rate on critical joints was 91.8% and the continuous hand-tracking rate 89.4%. Between 22 May and 5 August 2026 the platform recorded 409,527 alert events, with a mean uninterrupted tracking persistence of 72.0 s at 15 frames per second and a mean joint angle of 129.14° for high-risk events. Under approximately 250 concurrent requests the pipeline sustained 1.64 frames per second with about 650 ms end-to-end latency and no critical failures during a 152 s stress run, and the system passed formal acceptance testing with zero open issues (self-assessed Technology Readiness Level 6). The case shows that pose-based joint-angle alerting can be integrated into a live industrial IT environment, but also that the evidence produced does not yet establish ergonomic validity: no comparison with RULA/OWAS expert scoring, no joint-angle error against a reference measurement and no health-outcome data were obtained. These gaps define the agenda for moving from prototype to deployment.

Authors

Institutions

Publication Details

Journal
Journal of Computer Science and Technology Studies
Published
2026-11-20
DOI
https://doi.org/10.32996/jcsts.2026.8.18
Primary Topic
Ergonomics and Musculoskeletal Disorders
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

From Periodic Inspection to Continuous Monitoring: A Case Study of Camera-Based Joint-Angle Alerting for Ergonomic Risk on a Printed Circuit Board Assembly Line

Alessandra Castro Rocha, Daniel Angelo Oliveira de Abreu, Arthur Drumond Parente, Tony Douglas Martins Pena et al.
Journal of Computer Science and Technology Studies
Ergonomics and Musculoskeletal Disorders
article

From Periodic Inspection to Continuous Monitoring: A Case Study of Camera-Based Joint-Angle Alerting for Ergonomic Risk on a Printed Circuit Board Assembly Line

Alessandra Castro Rocha, Daniel Angelo Oliveira de Abreu, Arthur Drumond Parente, Tony Douglas Martins Pena, Tiago Carvalho de Almeida, Adriano Martins de Melo, Renato Moreira Teixeira Junior, Mariane Aguiar de Melo
article en

Abstract

Ergonomic risk management in electronics manufacturing still relies largely on manual, periodic inspections that offer little continuous visibility of operators' postural exposure. This case study reports the design, implementation and pilot-line validation of a Web-based ergonomic monitoring platform at an electronics manufacturing plant in Manaus, Brazil, developed over a seven-month applied research and development cycle (January–July 2026) by a technology institute in partnership with the manufacturer. The platform detects operators in industrial camera streams with a YOLOv8l detector, extracts body and hand landmarks with MediaPipe Pose and Hand Landmarker, computes joint angles, classifies each monitored joint into safe, light, moderate or high-risk angle bands with ordinal scores, and raises an alert only when a high-risk angle persists beyond a configurable holding time. Angle bands and persistence rules were defined with an NR-17-oriented ergonomics consultancy and are editable through the interface. Events flow through a containerised micro-service pipeline (RTSP ingestion, RabbitMQ messaging, REST API, SQL Server) to a Web interface with real-time monitoring, dashboards, PDF reports and a 3-D skeleton view. On a 500-frame sample from the surface-mount technology line, compared with human inspection, operator detection achieved 94.5% precision, 98.2% recall and a 96.3% F1-score; the reported hit rate on critical joints was 91.8% and the continuous hand-tracking rate 89.4%. Between 22 May and 5 August 2026 the platform recorded 409,527 alert events, with a mean uninterrupted tracking persistence of 72.0 s at 15 frames per second and a mean joint angle of 129.14° for high-risk events. Under approximately 250 concurrent requests the pipeline sustained 1.64 frames per second with about 650 ms end-to-end latency and no critical failures during a 152 s stress run, and the system passed formal acceptance testing with zero open issues (self-assessed Technology Readiness Level 6). The case shows that pose-based joint-angle alerting can be integrated into a live industrial IT environment, but also that the evidence produced does not yet establish ergonomic validity: no comparison with RULA/OWAS expert scoring, no joint-angle error against a reference measurement and no health-outcome data were obtained. These gaps define the agenda for moving from prototype to deployment.

Journal of Computer Science and Technology Studies
Universidade do Estado do Amazonas (BR), Instituto Federal do Amazonas (BR), Universidade Federal do Amazonas (BR)
Ergonomics and Musculoskeletal Disorders
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon