Intelligent PPE Monitoring with Computer Vision in Critical Infrastructure
Despite internal regulations and ongoing training, non-compliance was frequent. Some workers forgot part of their equipment; others partially removed it out of discomfort. This had led to several safety incidents, penalties from regulatory bodies, and a constant risk both to employees and to the facility's operational continuity.
Supervisors could not cover all access points simultaneously, and manual audits were insufficient to guarantee compliance in real time.
Real-time detection: We trained deep learning models capable of identifying each of the required PPE items: helmet, goggles, gloves, vest, footwear, and mask. The system analyzes every person approaching an access point in under 2 seconds.
Automated gate control: If the system detects that a worker is not wearing all the PPE required for that specific area, the access gate does not open. An auxiliary screen shows which item is missing, allowing the worker to correct the situation immediately.
Logging and traceability: Every access attempt is logged with a timestamp, worker identification (via badge), detected items, and outcome (access granted/denied). This provides full traceability for audits.
Supervision dashboard: Safety managers have a real-time panel with compliance statistics, alerts, and historical trends.
99.7% compliance: Within the first 3 months, PPE compliance rose from 72% to virtually total compliance.
Zero incidents due to missing protection: Since the system was deployed, no safety incident related to missing protective equipment has been recorded.
Elimination of regulatory penalties: The company successfully passed every safety audit conducted after the deployment.
Operational savings: The need for on-site supervisors at access points was reduced, reassigning that personnel to higher-value tasks.
Computer Vision (Deep Learning)
Automated Access Control
Real-Time Dashboard