Microsoft Azure AI Fundamentals (AI-900) Practice Exam

When creating an object detection model in the Custom Vision service, must you choose a class type?

1. Yes

2. No

3. Only for images with multiple objects

4. Only for animal detection

The correct answer is: No

In the context of using the Custom Vision service to create an object detection model, it is not mandatory to choose a class type when starting the project. You can set up an object detection model and train it without specifying predefined class types at the beginning. This flexibility allows for a more exploratory approach to developing models, where users can iteratively refine their training data and add classes as they improve their models over time. Choosing a class type can be beneficial for organization and clarity, especially when the project involves distinct categories of objects. However, it is not a strict requirement to start the model training process. Even with images containing multiple objects or focusing on specific categories like animals, one can begin the training without predetermined classes and later finalize or adjust them based on the data collected. This capability emphasizes the Custom Vision service's adaptability in accommodating various workflows and preferences, making it easier for developers to initiate their object detection tasks.