Amazon SageMaker

Input:

Following are the mandatory inputs for rotation correction algorithm on scanned documents:

• Supported content type: application/zip
• The algorithm corrects upside down pages in scanned documents.
• The algorithm works with scanned documents in formats – PDFs and Images. The input documents must be zipped.
• The input zip file can have up to 5 images [for types see below] or a scanned document in PDF format with maximum 5 pages.
• Also the image size must be less than 2 MB and PDF size must be less than 4 MB.
• Images can be of following types - bmp, dib, jpeg, jpg, jpe, png, pbm, pgm, ppm, tiff, tif

Output:

Instructions for score interpretation:

• Content type: application/zip
• Output zip file will contain Images / PDF with correct rotation depending on the documents inside Input zip file.
• Supported content types: ‘application/zip'

Invoking endpoint

AWS CLI Command

If you are using real time inferencing, please create the endpoint first and then use the following command to invoke it:

!aws sagemaker-runtime invoke-endpoint --endpoint-name $model_name --body fileb://$file_name --content-type 'application/zip' --region us-east-2 output.zip

Substitute the following parameters:

• "model-name" - name of the inference endpoint where the model is deployed
• file_name - input zip file name
• application/zip - type of the given input
• output.zip - filename where the inference results are written to

Resources:

• Sample Jupyter Notebook
• Sample Input File
• Sample Output File

Mphasis DeepInsights
Sample Input
Sample Jupyter Notebook