Corrosion Detection using Azure 8-week POC

The Corrosion Detection consulting solution helps to enable manufacturing and insurance industries to aid Quality Assurance team in identifying the right warranty tenure for metallic parts.

Our corrosion detection solution leverages Deep Learning based Convoluted Neural Networks to classify degree of corrosion in images based on probabilistic functions. The output of the Deep Learning Engine is consumed in the form of a web based online tool which is developed using REST API, helping the QA team in accurately identifying corrosion, reducing manual spot checks.

The solution leverages Azure services such as Azure IoT Hub, Azure ML Services, Azure Synapse Analytics, Azure Data Factory, Logic Apps etc.

Agenda (8-weeks):

• Discovery (Week 1)- Design solution implementation
• Development (Week 2-4)- Develop Pipelines, Train Model(s) & Model Validation(s)
• Holdout Data Validation (Week 4) - Validate models on completely unseen data
• Model Re-tweaking (Week 5-6) – Improve model on holdout data predictions
• Deployment (Week 7-8) - App Deployment/ ML pipeline deployment

Benefits:

• 80% accuracy in determining instance and degree of corrosions
• Brought down manual QC time by up to 65%
• Implemented the solution using an online tool through REST API

Why Affine