MicrosoftExpanded

Kyndryl automates defect detection on manufacturing lines with edge AI

Kyndryl deployed a custom defect detection solution in manufacturing environments to improve safety, reduce downtime, and streamline quality assurance. Using the Azure Percept edge AI development kit, Kyndryl built object detection models to monitor components on manufacturing lines in real-time. Images from production lines are processed at the edge, with results sent to Azure IoT Hub and analyzed with Azure Stream Analytics and Azure SQL Database. Visual insights are provided to plant managers via Power BI dashboards. This architecture reduced the need for manual inspections, accelerated response to quality issues, and enabled proactive maintenance. By integrating Azure Percept Studio for model training and deployment, the system quickly adapted to new product types, improving flexibility and operational efficiency.

Organization
Kyndryl
Location
India
Published
March 2022

Reported outcomes

Strategic outcomes

New product / capabilityDeployed real-time defect detection systemSpeed & agilityAccelerated response to quality issuesRisk & complianceImproved workplace safety and uptimeCost efficiencyReduced manual inspection effort and costs

Primary read

Use case focus

Showing 2 of 2

  • 1Automated Manufacturing Defect Detection with Edge AI
  • 2Real-time Component Counting and Quality Analytics
  • Manual defect detection was labor-intensive and error-prone.
  • Downtime and slow response to maintenance incidents increased operational costs.
  • Limited real-time monitoring decreased worker safety and responsiveness.
  • Implemented Azure Percept edge AI kit for on-site image capture and edge processing.
  • Trained custom computer vision object detection models in Azure Percept Studio.
  • Used Azure IoT Hub for collecting and relaying telemetry data to the cloud.
  • Analyzed data flows with Azure Stream Analytics and stored results in Azure SQL Database.
  • Enabled near-real-time data visualization using Power BI dashboards.
  • Significantly reduced manual inspection time and associated costs.
  • Improved real-time uptime and work safety.
  • Enabled proactive quality management and adaptive processes.
Architecture

The solution uses Azure Percept DK at the edge for vision-based defect detection, aggregating component counts via a custom module. Telemetry is relayed to Azure IoT Hub, transformed with Azure Stream Analytics, and stored in Azure SQL Database. Power BI visualizes results for plant managers and operators. All model training and deployment use Azure Percept Studio, and system operation is monitored and tuned via Azure IoT Explorer.

Sources & evidence1
ExpandedExpanded

The same organization appears in newer AI deployment evidence.

  • Same organization re-documented as recently as 2025.

Measures whether this deployment's public evidence persists — not whether the system is still in production.

Groundedness: Unavailable

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Was this useful?