Amazon MLS-C01 Exam Question: Labeling Accuracy for Video Surveillance Service

Machine Learning-Based Object Labeling for High Accuracy | AWS Service

Question

You are a machine learning specialist at a security firm that is building a video surveillance service to be used by police departments across the country.

This service needs to process the streaming video frames to find suspicious activity in public places such as train stations, subway platforms, etc.

To accomplish this task, your team needs to use a machine learning technique to find objects in the video frames on a list of objects identified as potentially dangerous, such as weapons.

You require to label your images by identifying the contents of your images at the pixel level for high accuracy. Which AWS service gives you the labeling accuracy your project requires?

Answers

Explanations

Click on the arrows to vote for the correct answer

A. B. C. D.

Answer: C.

Option A is incorrect.

Using the SageMaker Ground Truth Bounding Box labeling task, you can identify the pixel location of an object, but not identify the contents of an image at the pixel level.

Option B is incorrect.

Using the SageMaker Ground Truth Image Classification labeling task, your workers will classify your images using a predefined set of labels that you specify, but do not identify the contents of an image at the pixel level.

Option C is CORRECT.

Using the SageMaker Ground Truth Image Semantic Segmentation labeling task, your workers classify pixels in the image into a set of predefined labels or classes.

This will give you the pixel-level label identification accuracy you require.

Option D is incorrect.

The SageMaker Ground Truth Named Entity Recognition labeling task is used to extract information from unstructured text and classify it into predefined categories.

Reference:

Please see the Amazon SageMaker developer guide titled Use Amazon SageMaker Ground Truth to Label Data.

Please see the Amazon SageMaker developer guide titled Bounding Box.

Please see the Amazon SageMaker developer guide titled Image Semantic Segmentation.

Please see the Amazon SageMaker developer guide titled Image Classification (Single Label).

Please see the Amazon SageMaker developer guide titled Named Entity Recognition.

The AWS service that gives the labeling accuracy required for identifying the contents of images at the pixel level is the SageMaker Ground Truth Image Semantic Segmentation labeling task, option C.

Semantic segmentation is a technique in computer vision that involves labeling each pixel in an image with a corresponding class label. This technique allows for more accurate object detection and recognition in images and videos.

SageMaker Ground Truth is a managed data labeling service provided by AWS that makes it easy to label datasets using human annotators or machine learning algorithms. Ground Truth provides pre-built templates for common labeling tasks, including image classification, object detection, and semantic segmentation.

Option A, SageMaker Ground Truth Bounding Box labeling task, is used for object detection tasks that require drawing bounding boxes around objects of interest. This is useful for identifying the location of objects in an image, but it does not provide the pixel-level accuracy required for semantic segmentation.

Option B, SageMaker Ground Truth Image Classification labeling task, is used for identifying the presence or absence of specific classes in an image. This is useful for tasks such as identifying whether an image contains a specific object, but it does not provide the pixel-level accuracy required for semantic segmentation.

Option D, SageMaker Ground Truth Named Entity Recognition labeling task, is used for identifying and classifying named entities such as people, organizations, and locations in text data. This is not relevant for the video surveillance service described in the question.

Therefore, the correct option is C, SageMaker Ground Truth Image Semantic Segmentation labeling task, which provides the pixel-level accuracy required for identifying the contents of images in the video surveillance service.