Developing Custom Vision Solutions

What all the approaches to use the Custom Vision service to train an image classification model?

• The Custom Vision portal, the Custom Vision REST API or SDK, or a combination of both approaches.

• Only the Custom Vision portal

• The Custom Vision API or SDK, and the Azure Machine Learning portal

• The Azure portal or the Custom Vision portal

To use the Custom Vision service, you must provision which two kinds of Azure resources?

• A training resource

• An image resource

• An Azure AI resource

• A prediction resource

If a model has been trained to allow each image to be associated with multiple labels, what type of classification is that considered?

• Multiclass classification

• Multi-image classification

• Multilabel classification

• Multi-object classification

How does Object Detection differ from Image Classification?

• Object Detection uses bounding boxes around objects within an image, while Image Classification will apply a label to the entire image.

• Image Classification uses bounding boxes around objects within an image, while Object Detection will apply a label to the entire image

• Object Detection can only identify objects based on an input model you provide while Image Classification uses complex AI to determine the objects.

• Image Classification can only identify objects based on an input model you provide while Object Detection uses complex AI to determine the objects

When labeling images in the Custom Vision portal, which statement is true?

• You must tag every single image.

• You can tag an initial batch of images and then use the smart labeler tool.

• You must tag every single image, but then use the smart labeler tool to define the bounding box.

• You can tag an initial batch of images and use the smart labeler tool, but must specify the bounding box

What is the difference between a model trained for multiclass classification versus multilabel classification?

Answer explanation

Your Answer: Models can be trained for multiclass classification (in other words, there are multiple classes, but each image can belong to only one class) or multilabel classification (in other words, an image might be associated with multiple labels)

What is the class label for an image classification project?

Answer explanation

Your Answer: Image classification is a computer vision technique in which a model is trained to predict a class label for an image based on its contents. Usually, the class label relates to the main subject of the image.

What is an example of Object Detection used at an in-person store?

Answer explanation

Your Answer: An AI-enabled checkout system in a grocery store might need to identify the type and location of items being purchased by a customer.

What is the most significant difference between training an image classification model and training an object detection model?

Answer explanation

Your Answer: While image classification requires one or more tags that apply to the whole image, object detection requires that each label consists of a tag and a region that defines the bounding box for each object in an image. The Custom Vision portal provides a graphical interface that you can use to label your training images.

AminaTech wants to publish an app that allows customers to take a picture of individual pieces of fruit and the app will identify the type of fruit. Describe how you would use the Custom Vision portal to do this.

Answer explanation

Your Answer: Should include a description of using the portal to upload images of fruit, tag the images, train and evaluate a classification model, test the model, and then publish the trained model to a prediction resource that can be called from an API.