AI-102 Question 142
Select 4You are tasked with building a custom image classification model using Azure Cognitive Services Custom Vision. Your goal is to classify images of different types of fruits (e.g., apples, oranges, bananas). You have already collected a dataset of labeled fruit images. Which of the following steps are necessary to successfully train and deploy the custom image classification model?
- A
Create a Custom Vision project and select a classification domain suitable for your dataset.
- B
Upload labeled images to the Custom Vision project and tag them with the appropriate class labels.
- C
Manually write the classification algorithm in Python after uploading the dataset.
- D
Train the model in the Custom Vision portal or using the SDK, and evaluate its performance on a validation set.
- E
Export the trained model and integrate it into your application or deploy it as an endpoint.
Show answer and explanation
Correct answers: A, B, D, E
Explanation
To train and deploy a custom image classification model using Azure Cognitive Services Custom Vision, you must create a project, upload labeled data, train the model, evaluate its performance, and deploy or export the model. Writing a manual algorithm is not required, as the Custom Vision service handles the training process for you.
- A. Correct.
Correct: Creating a Custom Vision project and selecting an appropriate classification domain is a key step to ensure the model is trained with the right settings and capabilities.
- B. Correct.
Correct: Uploading labeled images and tagging them ensures the model learns from correctly categorized data, which is essential for training.
- C. Incorrect.
Incorrect: You do not need to manually write a classification algorithm in Python when using the Custom Vision service. The platform automates this process for you.
- D. Correct.
Correct: Training the model and evaluating its performance ensures the model meets accuracy requirements before deployment.
- E. Correct.
Correct: Exporting the model or deploying it as an endpoint is necessary to make it accessible for integration into your application.