AWS Rekognition

Learn how to configure AWS Rekognition for use with Filebase.

What is AWS Rekognition?

AWS Rekognition is is a software service hosted by Amazon that provides machine learning image and video analysis. AWS Rekognition can provide feedback regarding image and video content, attributes, and components for use-cases such as racial recognition and other machine learning algorithms.

Read below to learn how to use AWS Rekognition with Filebase.

Prerequisites:

• Have a valid AWS account.

• Sign up for a free Filebase account.

• Have your Filebase Access and Secret Keys. Learn how to view your access keys here.

• Create a public Filebase Bucket. Learn how to create a bucket here. Note: This feature requires a paid Filebase subscription.

• Download, install and configure the AWS CLI.

1. Navigate to the Filebase web console dashboard. Select the Buckets menu.

2. Select a Public Filebase bucket, or create one if necessary.

Note: This feature requires a paid Filebase subscription.

3. Upload an image file to your public bucket.

4. Once uploaded, select the object to view its details.

5. Copy the Object URL to the image file.

6. Next, create a .JSON file on your local environment with the following content:

{
"CORSRules":[
    {
        "AllowedHeaders": [],
        "AllowedMethods": [
            "GET"
        ],
        "AllowedOrigins": [
            "*"
        ],
        "ExposeHeaders": []
    }
    ]
}

7. Save this file as corspolicy.json.

8. Open a terminal window and run the following command to apply the CORS policy to your Filebase bucket:

aws --endpoint https://s3.filebase.com s3api put-bucket-cors --bucket BUCKET-NAME --cors-configuration=file://corspolicy.json

Replace BUCKET-NAME with the name of your public Filebase bucket.

9. Navigate to the AWS Rekognition Console. Input the Filebase object URL under the ‘Use your own image’ option. Then select ‘Go’.

10. Once uploaded, AWS Rekognition will return information regarding the content of the photo.

If you have any questions, please join our Discord server, or send us an email at [e-mail]