You work as a machine learning specialist for a software company that is building a bird identifier mobile app.
The app will allow users to take a photo of a bird and then use the app to process the image and identify the bird.
You are building your data repository for your machine learning model to use for training.
You need to label the very large set of images you plan to use for training.
You have decided to use SageMaker Ground Truth to label your images.
You are creating the Ground Truth worker tasks for your workers to use when performing their labeling tasks.
You need to identify the birds in your images at the pixel level for accuracy.
Which SageMaker Ground Truth built-in task should you use?
Click on the arrows to vote for the correct answer
A. B. C. D.Correct Answer: D.
Option A is incorrect.
The Bounding Box SageMaker Ground Truth built-in task is used to allow your workers to use a box that bounds various subjects in the image.
You need your workers to identify the subjects (birds) at the pixel level.
Option B is incorrect.
The Image Classification (Single Label) SageMaker Ground Truth built-in task is used to allow your workers to label subjects in your images using a predefined set of labels provided by you.
You need your workers to identify the subjects (birds) at the pixel level.
Option C is incorrect.
The Image Classification (Multi-Label) SageMaker Ground Truth built-in task is used to allow your workers to label multiple subjects in your images.
You need your workers to identify the subjects (birds) at the pixel level.
Option D is correct.
The Semantic Segmentation Ground Truth built-in task is used to allow your workers to label subjects in your image at the pixel level.
Reference:
Please see the Amazon SageMaker developer guide titled Label Data (https://docs.aws.amazon.com/sagemaker/latest/dg/data-label.html), the Amazon SageMaker developer guide titled Built-in Task Types (https://docs.aws.amazon.com/sagemaker/latest/dg/sms-task-types.html), the Amazon SageMaker developer guide titled Image Classification (Single Label) (https://docs.aws.amazon.com/sagemaker/latest/dg/sms-image-classification.html), the Amazon SageMaker developer guide titled Image Classification (Multi-label) (https://docs.aws.amazon.com/sagemaker/latest/dg/sms-image-classification-multilabel.html), and the Amazon SageMaker developer guide titled Image Semantic Segmentation (https://docs.aws.amazon.com/sagemaker/latest/dg/sms-semantic-segmentation.html), and the Amazon SageMaker developer guide titled Bounding Box (https://docs.aws.amazon.com/sagemaker/latest/dg/sms-bounding-box.html)
For the given scenario, the appropriate SageMaker Ground Truth built-in task to use is Image Semantic Segmentation.
Image Semantic Segmentation is a built-in task in Amazon SageMaker Ground Truth that allows you to label images at a pixel level. This task involves segmenting an image into regions and assigning labels to each region. This task is particularly useful when you want to identify objects or regions of interest within an image, as in the case of bird identification in the given scenario.
The other three built-in tasks provided by SageMaker Ground Truth are:
A. Bounding Box: This task involves drawing a rectangle around an object of interest in an image. This task is useful when you want to locate an object in an image, but it does not provide the pixel-level accuracy required for identifying birds in the given scenario.
B. Image Classification (Single Label): This task involves assigning a single label to an entire image. This task is useful when you want to classify images into broad categories, but it does not provide the pixel-level accuracy required for identifying birds in the given scenario.
C. Image Classification (Multi-Label): This task involves assigning multiple labels to an image. This task is useful when you want to classify images into multiple categories, but it does not provide the pixel-level accuracy required for identifying birds in the given scenario.
Therefore, Image Semantic Segmentation is the most appropriate built-in task to use for the given scenario.