Microsoft Certified: Azure AI Engineer AssociateImplement image and video processing solutionsHard
A security firm is developing an automated surveillance system. The system needs to detect instances where specific, unauthorized objects (e.g., prohibited weapons, suspicious packages) appear in real-time video feeds. The objects are unique and not part of any standard object detection dataset. The solution must provide high accuracy and low latency for immediate alerts. Which Azure service, configured as an object detection model, should be utilized, and what type of deployment strategy would ensure low latency for real-time video analysis?
- AAzure Computer Vision, deployed as a cloud-based API endpoint.
- BAzure Custom Vision, deployed as an IoT Edge module.
- CAzure Video Indexer, with its built-in object recognition.
- DAzure Face API, deployed with a custom face list.
Show answer & explanationAnswer & explanation
Correct answer: B. Azure Custom Vision, deployed as an IoT Edge module.
Azure Custom Vision allows training highly accurate custom object detection models for unique objects like unauthorized weapons. For low latency and real-time video analysis, especially in surveillance, deploying the trained Custom Vision model as an IoT Edge module allows inference to occur directly on the edge device, minimizing network latency and bandwidth usage.
Why the other options are wrong
- A. Computer Vision's pre-trained models won't detect 'specific, unauthorized objects'. Cloud deployment adds latency.
- C. Video Indexer is for broader video insights, not custom object detection with low-latency edge deployment for specific unauthorized items.
- D. Face API is for faces, not general object detection. Custom face lists are irrelevant here.
Custom Vision IoT Edge Deployment
Deploying an Azure Custom Vision model as an IoT Edge module enables on-device inference for custom object detection, providing low-latency real-time analysis for scenarios like surveillance or manufacturing line monitoring.
- Custom Vision for specific object detection.
- IoT Edge for on-device, low-latency inference.
- Ideal for real-time video and image streams.
Memory trick: Custom eyes on the edge, for immediate insights and no latency hedge.