Microsoft Azure AI Fundamentals (AI-900) Practice Exam

Can you select from predefined domains when creating an object detection model in the Custom Vision service?

1. Yes

2. No

3. Only if using custom datasets

4. Only if choosing multiclass type

The correct answer is: Yes

When creating an object detection model in the Custom Vision service, users can indeed select from predefined domains. This feature allows for leveraging models that have been pre-trained on specific types of images, which can significantly enhance the performance and efficiency of the object detection process. By choosing a predefined domain, the model benefits from a foundational knowledge base that reflects particular characteristics or contexts, making the training process more effective. In scenarios where a well-defined set of labels is required and where the images resemble those found in the predefined domains, starting with these options can streamline the model development. It allows users to capitalize on existing research and development in the field of computer vision, ultimately saving time and resources while increasing accuracy. This ability is particularly useful because it helps users build models that are more effective, especially if they may not have extensive datasets or specific expertise in training object detection models from scratch.