Creating a manifest file

You can create a dataset by importing an SageMaker Ground Truth format manifest file. If your images are labeled in a format that isn't a SageMaker Ground Truth manifest file, use the following information to create an SageMaker Ground Truth format manifest file.

Manifest files are in JSON lines format where each line is a complete JSON object representing the labeling information for an image. There are different formats for image classification and image segmentation. Manifest files must be encoded using UTF-8 encoding.

Note

The JSON line examples in this section are formatted for readability.

The images referenced by a manifest file must be located in the same Amazon S3 bucket. The manifest file can be in a different bucket. You specify the location of an image in the source-ref field of a JSON line.

You can create a manifest file by using code. The Amazon Lookout for Vision Lab Python Notebook shows how to create an image classification manifest file for the circuitboard example images. Alternatively, you can use the Datasets example code in the AWS Code Examples Repository. You can easily create a manifest file by using a Comma Separated Values (CSV) file. For more information, see Creating a classification manifest file from a CSV file.

Topics

• Defining JSON lines for image classification
• Defining JSON lines for image segmentation
• Creating a classification manifest file from a CSV file
• Creating a dataset with a manifest file (console)
• Creating a dataset with a manifest file (SDK)