Creating a dataset using images stored in an Amazon S3 bucket

You can create a dataset using images stored in an Amazon S3 bucket. With this option, you can use the folder structure in your Amazon S3 bucket to automatically classify your images. You can store the images in the console bucket or another Amazon S3 bucket in your account.

Setting up folders for automatic labeling

During dataset creation, you can choose to assign label names to images based on the name of the folder that contains the images. The folders must be a child of the Amazon S3 folder path that you specify in S3 URI when you create the dataset.

The following is the train folder for the Getting Started example images. If you specify the Amazon S3 folder location as S3-bucket/circuitboard/train/, the images in the folder normal are assigned the label Normal. Images in the folder anomaly are assigned the label Anomaly. The names of deeper child folders aren't used to label images.

S3-bucket
  └── circuitboard
    └── train
        ├── anomaly
            ├── train-anomaly_1.jpg
            ├── train-anomaly_2.jpg
            ├── .
            └── .
        └── normal
            ├── train-normal_1.jpg
            ├── train-normal_2.jpg
            ├── .
            └── .
 

Creating a dataset using images from an Amazon S3 bucket

The following procedure creates a dataset using the classification example images stored in an Amazon S3 bucket. To use your own images, create the folder structure described in Setting up folders for automatic labeling.

The procedure also shows how to create a single dataset project, or a project that uses separate training and test datasets.

If you don't choose to automatically label your images, you need to label the images after the datasets is created. For more information, see Classifying images (console).

Note

If you've just completed Creating your project, the console should show your project dashboard and you don't need to do steps 1 - 4.

To create a dataset using images stored in an Amazon S3 bucket

1. If you haven't already done so, upload the getting started images to your Amazon S3 bucket. For more information, see Image classification dataset.

2. Open the Amazon Lookout for Vision console at https://console.aws.amazon.com/lookoutvision/.

3. In the left navigation pane, choose Projects.

4. In the Projects page, choose the project to which you want to add a dataset. The details page for your project is displayed.

5. Choose Create dataset. The Create dataset page is shown.

   Tip

   If you're following the Getting Started instructions, choose Create a training dataset and a test dataset.

6. Choose the Single dataset tab or the Separate training and test datasets tab and follow the steps.

   Single dataset

   1. In the Dataset configuration section, choose Create a single dataset.

   2. Enter the information for steps 7 - 9 in the Image source configuration section.

   Separate training and test datasets

   1. In the Dataset configuration section, choose Create a training dataset and a test dataset.

   2. For your training dataset, enter the information for steps 7 - 9 in the Training dataset details section.

   3. For your test dataset, enter the information for steps 7 - 9 in the Test dataset details section.

      Note

      Your training and test datasets can have different image sources.

7. Choose Import images from Amazon S3 bucket.

8. In S3 URI, enter the Amazon S3 bucket location and folder path. Change bucket to the name of your Amazon S3 bucket.

   1. If you're creating a single dataset project or a training dataset, enter the following:

      s3://bucket/circuitboard/train/

   2. If you're creating a test dataset enter the following:

      s3://bucket/circuitboard/test/

9. Choose Automatically attach labels to images based on the folder.

10. Choose Create dataset. A dataset page opens with your labeled images.

11. Follow the steps in Training your model to train your model.