You want to draw pixel level labels for your images so you can create a dataset for training your machine learning models.
Can you do this by setting up a semantic segmentation job in SageMaker Ground Truth?
--
SageMaker is a fully managed service to prepare data and build, train, and deploy machine learning (ML) models for any use case with fully managed infrastructure, tools, and workflows.
--
Tutorial Summary:
- Navigate to Amazon S3 service to upload images of cats and dogs for labeling.
- Access SageMaker service, go to Ground Truth, and create a labeling job.
- Choose automated data setup, specifying the S3 bucket and data type as images.
- Create a new IM role, granting permissions to the S3 bucket.
- Confirm the creation of a new manifest file in the S3 bucket and proceed to select the task type as 'image' in the task type section.
- Choose 'semantic segmentation' for drawing pixel-level labels.
- Choose private workers, enter team details, and define labeling tools (task timeout, expiration, organization name, support email).
- Enter task descriptions, labels (e.g., 'cat', 'dog'), and confirm using the preview.
- Access the labeling job link from the inbox, log in with credentials, and start labeling by drawing polygons around the cat and dog labels in the images.
- Submit the labeled images and wait for the labeling job status to change to complete.
- Review the output, confirming the presence of green and blue polygons around labeled objects.
- Download the output manifest file from the output folder in the S3 bucket to view technical details of the semantic segmentation job.
Негізгі бет AWS SageMaker GroundTruth 👉 Label Job Semantic Segmentation Tutorial 🔥
Пікірлер: 8