Azure Computer Vision 3.0 PDF Reading

[Alexander.Duff]

Afternoon,

I'm trying to pass through Binary images (in a collection) through the Azure Computer Vision 3.0 asset and have an issue.

BP Version: 6.0
Asset: Microsoft Computer Vision v3.0
Object: OCR Raw Image & Get Read Results

Issue
After altering
From: $.readResults

To: $.analyzeResult.readResults​

to record the results, I'm presented with a collection of analysed results.

This collection contains 2 rows. The issue here is that it is a 4 page PDF (First-page front and back/Second-page front and back) So what this means is that I should have 4 rows. 2 rows containing texts and 2 rows that should be blank.

Is there any way I can force read all pages? I'm not 100% why it's only pulling back 2 pages when there're clearly 4 pages in the pdf.

[ewilson]

Hi @Alex Duff,

So you're passing in images you created of the 4 page PDF (4 images), correct? What image format are you using?

Here's an excerpt from the Azure Vision reference that describes image requirements:


Do all 4 of the images meet the above requirements?

Cheers,
Eric​

[ewilson · Accepted solution]

FYI - There's also a note on the API reference page for the Read action that if you're using the Azure Free Tier, they will only process the first 2 pages.

Cheers,
Eric