The Challenge:

Enforcing Health, Safety and Environment (HSE) protocols on drilling platforms is one complicated task faced by the Oil & Gas Industry. Tracking and identifying staff and their behaviors is often complex and time-consuming. Even more so is monitoring safety procedures and guaranteeing Personal Protective Equipment (PPE) compliance.


If disregarded, these crucial operational safe measures can impact Organizational KPIs, increase risk, create liability, and result in higher costs. Adopting technology that automates HSE monitoring, and triggers timely actionable alerts when detecting infringements to safety protocols, could drastically reduce risks resulting in substantial savings on HSE.

The Solution:

Insight-AI was adopted for monitoring operations via automatic footage annotation on CCTV streams in drilling facilities. By identifying and classifying machinery, personnel, and behaviors, the system is able to trigger actionable real-time alerts that reflect qualitative and quantitative degrees of violation of safety protocols, and indexed across the following dimensions:

PPE compliance: presence and usage of PPE wear and accessories, such as helmets, gloves, boots, and goggles.

Safety areas positioning: appropriate and safe spatial distribution of personnel across the work areas.

Accidents detection: occurrence of sudden potentially catastrophic scenarios, such as falls, rapid unscheduled disassembly, paranormative deambulation.

Forbidden behavior: identification of conducts that violate safety principles and rules, such as carrying concealed incendiary devices, or engaging in high risk physical activity.

The Technology:

Insight-AI performs automated information extraction on CCTV video streams by joining different layers of video analytics. This enables heterogeneous neural networks to be applied either in series or in parallel, in order to more combine information from each of them into actionable insights, within three macro areas:

Space Operationalization: in order to gain contextual awareness of a custom area of interest (AOI), Insight-AI algorithms collocates and contextualizes footage of each camera into a “blueprint” parent visual scene.

Object and Pose Detection: each footage annotation layer is trained to recognize and identify a subset of target personnel, machinery, and equipment: a modular architecture guarantees that models with increasing specificity and sensitivity are progressively trained and served, according to work in progress requirements.

Behavior Detection: by combining spatial awareness with object and personnel recognition capabilities, indexes high order behavior can be extracted from each and every video stream, enabling the system to provide a wide spectrum information, which can be tailored on a camera by camera basis.


Explore Datumcon Products

Neural Networks for Video, Language, and Geospatial Analytics.

Schedule a demo

Discover more about Neural Networks applied to HSE on drilling platforms.