What challenges did AWS customers in upstream oil and gas industry face before AI implementation?

Remote well pads require effective safety and logistics monitoring but suffer from limited connectivity and power constraints. Verifying vendor truck visits and PPE usage is difficult with manual or disconnected processes.

What was the impact of the AI solution?

Operates with minimal data transmission to cloud, suitable for limited connectivity environments. Estimated cost of $161 per month to run at 1,000 well pads excluding hardware costs.

Source

ProductionEvidence: Low40/100

Improving Safety and Logistics at Well Pads with AWS AI Computer Vision

Use case typeIndustrial inspectionUpdated Jun 13, 2026

AWS customers in the upstream oil and gas industry faced challenges monitoring safety and logistics at remote well pads with limited connectivity. They deployed an AI solution using AWS DeepLens cameras running custom ML models locally to detect trucks, license plates, and PPE compliance such as hard hats. The solution leverages Amazon Rekognition for image analysis, Amazon Textract for extracting vehicle markings, and AWS IoT Greengrass to manage edge runtime and connectivity. Processed images and alerts are sent to the cloud via AWS IoT Core, integrated with Amazon Aurora for data storage, Amazon QuickSight for dashboards and visualization, Amazon SNS for notifications, and Amazon CloudWatch for logging and monitoring. This enables real-time validation of vendor truck visits and PPE compliance, reduces operational costs, and improves safety monitoring without continuous cloud connectivity.

Organization: AWS customers in upstream oil and gas industry
Industry: Energy & Utilities
Location: United States
Published: August 2021

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: AWS customers in upstream oil and gas industry
Provider: AWS
Maturity: Production
Linked source: AWS Industry Blog

They deployed an AI solution using AWS DeepLens cameras running custom ML models locally to detect trucks, license plates, and PPE compliance such as hard hats

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Edge AI
2Safety Monitoring
3Computer Vision

Deployed custom computer vision models on AWS DeepLens edge cameras to detect trucks, license plates, and PPE compliance.
Used Amazon Rekognition and Amazon Textract for image processing and object recognition in the cloud.
Integrated edge inference with cloud services using AWS IoT Greengrass, AWS IoT Core, and Lambda for event processing and alerting.
Stored and analyzed data with Amazon Aurora and visualized results with Amazon QuickSight dashboards.
Monitored system health and generated alerts via Amazon CloudWatch and Amazon SNS.

Technologies

AWS DeepLens Amazon Rekognition Amazon Textract AWS IoT Greengrass AWS IoT Core AWS Lambda Amazon Aurora Amazon QuickSight Amazon SNS Amazon CloudWatch

Operates with minimal data transmission to cloud, suitable for limited connectivity environments.
Estimated cost of $161 per month to run at 1,000 well pads excluding hardware costs.

Architecture

The solution architecture includes AWS DeepLens running ML models on cameras at the edge to perform object detection and PPE compliance checks locally. Images with objects of interest are sent via AWS IoT Core to cloud services. Amazon Rekognition and Amazon Textract perform image analysis and text extraction to identify vehicle and vendor information. Data is stored in Amazon Aurora to support dashboards and reporting via Amazon QuickSight. AWS Lambda and Amazon SNS handle event processing and notifications. Amazon CloudWatch monitors system health and logs.

Sources & evidence1

Evidence: Low40/100Evidence strength