MLS-C01 Question 92
Select 3You are tasked with building a machine learning model to classify images of damaged vehicles into specific damage categories (e.g., 'broken windshield', 'dented door'). Your dataset contains thousands of unlabeled images. To create a high-quality labeled dataset efficiently, you plan to use Amazon SageMaker Ground Truth with Amazon Mechanical Turk as the workforce. Which of the following steps are required to set up and execute this labeling process?
- A
Define a labeling job in Amazon SageMaker Ground Truth and specify the input dataset location in Amazon S3.
- B
Choose Amazon Mechanical Turk as the labeling workforce in the SageMaker Ground Truth console.
- C
Manually distribute the images among Amazon Mechanical Turk workers via email.
- D
Set up labeling task instructions, including detailed guidelines for labeling and example labels.
- E
Create a Lambda function to automatically evaluate the labeled data for quality control.
Show answer and explanation
Correct answers: A, B, D
Explanation
To use Amazon SageMaker Ground Truth with Amazon Mechanical Turk, you must first define a labeling job in SageMaker, specify the input dataset location, select Amazon Mechanical Turk as the workforce, and provide clear task instructions for workers. SageMaker Ground Truth automates task distribution and provides built-in quality control tools, so manual task distribution or custom Lambda functions are not required.
- A. Correct.
Correct. Defining a labeling job and specifying the input dataset location in Amazon S3 is a key step to initiate the labeling process in Amazon SageMaker Ground Truth.
- B. Correct.
Correct. Amazon Mechanical Turk can be chosen as the workforce in SageMaker Ground Truth to crowdsource labeling tasks.
- C. Incorrect.
Incorrect. Manually distributing images among workers via email is not required or supported. SageMaker Ground Truth handles the distribution of tasks automatically.
- D. Correct.
Correct. Providing detailed labeling instructions and examples is critical for ensuring labeling consistency and accuracy.
- E. Incorrect.
Incorrect. While quality control mechanisms are important, SageMaker Ground Truth includes built-in features such as automated data quality checks, so creating a separate Lambda function for this is unnecessary.