Artificial Intelligence (AI) Based Personnel Protective Equipment (PPE) Monitoring – Case Study in Rokan Drilling Operation

Abstract

Automatic Personal Protective Equipment (PPE) Monitoring System is a term for an image processing method which is used to identify compliance to PPE utilization practices by personnel on the field. The vast diversity and assortment of PPEs across companies make it quite challenging to take existing limited public dataset models from other companies and apply them directly to specific companies such as Pertamina Hulu Rokan. As a state-owned company that operates one of the largest fields in Indonesia with an extensive drilling program, Pertamina Hulu Rokan decided to use an Automatic PPE monitoring system with online Closed-Circuit Television (CCTV) units to enable recognition of PPE objects in drilling operations by the implementation of Artificial Intelligence. As a first step towards building this system, we proposed to build PPE datasets from various Rig areas and in different light conditions. The next step was to use deep learning technology such as Yolov4 and train the model using the PPE datasets to localize PPE objects used by personnel such as gloves, gloves-off, helmets, helmets-off, shoes, shoes-off, masks, masks-off, glasses, glasses-off. Our Preliminary results indicate that our method has been useful to identify compliance of PPE usage during operation work and to minimize the safety risk exposure of our personnel.