Microsoft Certified: Azure AI Engineer AssociateImplement image and video processing solutionsMedium
A manufacturing plant is implementing an automated visual inspection system. They need to identify subtle defects on product surfaces, such as small scratches or discolorations, which are unique to their product line and not easily detectable by general-purpose vision models. Which approach should they use?
- ATrain an Azure Custom Vision classification model.
- BApply Azure Form Recognizer for defect detection.
- CImplement Azure Video Indexer for defect analysis.
- DUse Azure Computer Vision's pre-trained object detection.
Show answer & explanationAnswer & explanation
Correct answer: A. Train an Azure Custom Vision classification model.
Azure Custom Vision allows training a custom image classification model with specific labeled images of both good and defective products. This enables the model to learn and identify subtle, unique defects that pre-trained general models cannot.
Why the other options are wrong
- B. Form Recognizer is for extracting structured data from documents, unrelated to visual defect detection on products.
- C. Video Indexer is for analyzing video content for insights like speech, faces, and topics, not for surface defect detection on static images.
- D. Pre-trained object detection models are for common objects and would not recognize subtle, specific product defects without custom training.
Custom Vision Classification
A feature of Azure Custom Vision that allows users to train models to categorize images into custom-defined classes based on their content.
- Requires labeled images for training
- Outputs a class label for an input image
- Ideal for identifying specific types of defects or conditions
- Can be deployed as an endpoint or exported for edge devices
Memory trick: Custom Vision Classifies Subtle Scratches.