What challenges did Grid Dynamics face before AI implementation?

Improving quality control processes by automating defect detection on assembly and sorting lines to reduce quality costs and defect rates.

What was the impact of the AI solution?

Achieved 85% recall with 95% precision for object detection and 100% precision and recall for anomaly classification on test data. The solution provides real-time anomaly detection suitable for production-grade deployment with continuous training to improve over time.

Source

GCPEvidence: Medium50/100

Visual Quality Control in Manufacturing with Google Cloud Vertex AI by Grid Dynamics

Use case typeComputer vision inspectionUpdated Feb 14, 2023

Grid Dynamics described a real-time visual quality control solution for defect detection on assembly and sorting lines. The pipeline consumes video streams from cameras, identifies parcels with Vertex AI AutoML, classifies anomalies, and uses a custom tracking algorithm to monitor objects in real time.

Organization: Grid Dynamics
Industry: Manufacturing
Location: United States
Published: February 2023

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Object detection recall: 85% increase

Grid Dynamics blogFeb 14, 2023Blog postExplicit claimMedium evidence strength

For object detection, the model achieves 85% recall with 95% precision.

Normalized claim

Object detection precision: 95% increase

Grid Dynamics blogFeb 14, 2023Blog postExplicit claimMedium evidence strength

For object detection, the model achieves 85% recall with 95% precision.

Normalized claim

Anomaly classification precision: 100% increase

Grid Dynamics blogFeb 14, 2023Blog postExplicit claimMedium evidence strength

Next, we achieved a precision and recall score of 100% for object classification.

Normalized claim

Anomaly classification recall: 100% increase

Grid Dynamics blogFeb 14, 2023Blog postExplicit claimMedium evidence strength

Next, we achieved a precision and recall score of 100% for object classification.

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

Customer: Grid Dynamics
Provider: GCP
Maturity: Unknown
Linked source: Grid Dynamics blog

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 2 of 2

1Computer vision inspection
2Quality management

Developed a real-time visual quality control pipeline using Vertex AI and AutoML services for object detection and anomaly classification.
The system uses a camera video stream, manual labelling and augmentation for a small dataset, and a custom multi-object tracking algorithm to support real-time monitoring.

Technologies

Vertex AI Vertex AI AutoML Google Cloud Platform

Achieved 85% recall with 95% precision for object detection and 100% precision and recall for anomaly classification on test data.
The solution provides real-time anomaly detection suitable for production-grade deployment with continuous training to improve over time.

Sources & evidence1

Evidence: Medium50/100Evidence strength