AI-102 Question 144
Select 4You are tasked with building a custom image classification model to identify different types of fruits (e.g., apples, bananas, oranges) using Azure AI. You want to use Azure Custom Vision to train the model. Which steps should you take to successfully train and publish the model?
- A
Upload a labeled image dataset into Azure Custom Vision.
- B
Choose the 'Object Detection' domain in Azure Custom Vision for this scenario.
- C
Split the dataset into training and testing datasets to evaluate model performance.
- D
Train the model using the labeled dataset in Azure Custom Vision.
- E
Export the trained model as a TensorFlow file for deployment.
- F
Publish the trained model to generate a prediction endpoint.
Show answer and explanation
Correct answers: A, C, D, F
Explanation
To successfully train and deploy a custom image classification model using Azure Custom Vision, you need to upload a labeled dataset, split it into training and testing datasets for evaluation, train the model, and publish it to generate a prediction endpoint. The 'Object Detection' domain is not suitable for classification tasks, and exporting the model is optional unless specifically required for custom deployment scenarios.
- A. Correct.
Correct: Uploading a labeled dataset is essential for training the model. Labeled images provide the data required to teach the model to classify images.
- B. Incorrect.
Incorrect: The 'Object Detection' domain is not appropriate in this case, as the requirement is to classify the entire image into specific fruit categories. The 'Image Classification' domain should be used instead.
- C. Correct.
Correct: Splitting the dataset into training and testing datasets is a best practice to ensure you can evaluate your model's performance on unseen data.
- D. Correct.
Correct: Training the model with the labeled dataset is a necessary step to allow the model to learn how to classify images.
- E. Incorrect.
Incorrect: Exporting the model as a TensorFlow file is optional and not required for deploying the model via Azure Custom Vision. Deployment can be done using the prediction endpoint generated after publishing the model.
- F. Correct.
Correct: Publishing the trained model generates a prediction endpoint, which can be used to classify new images.