search

MicrosoftDocs / azure-docs

Open source documentation of Microsoft Azure
https://docs.microsoft.com/azure
Creative Commons Attribution 4.0 International
10.26k stars 21.43k forks source link

V3 OCR Error? #71404

Closed JohnStraumann66 closed 3 years ago

JohnStraumann66 commented 3 years ago

Hello:

With the previous version of the Vision OLCR service, the text in the attached image was recognized perfectly. Now with teh V3 version, this is what I get:

Ano{he.r angry alun de an a,slouning number of aardvarb OCR2

Any ideas?

Thanks!

John.

[Enter feedback here]

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

YutongTie-MSFT commented 3 years ago

Thanks for the feedback! We are currently investigating and will update you shortly.

YutongTie-MSFT commented 3 years ago

Thanks a lot for the feedback, I have forwarded this to product team for investigating and fixing. I will recommend you to move to Read API. The Read API is optimized for text-heavy images and multi-page, mixed language, and mixed type (print – seven languages and handwritten – English only) documents.

https://westus.dev.cognitive.microsoft.com/docs/services/computer-vision-v3-1-ga/operations/5d986960601faab4bf452005

Below is the result I got from Read API.

{
  "status": "succeeded",
  "createdDateTime": "2021-03-02T22:07:25Z",
  "lastUpdatedDateTime": "2021-03-02T22:07:26Z",
  "analyzeResult": {
    "version": "3.0.0",
    "readResults": [{
      "page": 1,
      "angle": 0,
      "width": 338,
      "height": 138,
      "unit": "pixel",
      "language": "en",
      "lines": [{
        "boundingBox": [11, 13, 279, 14, 279, 38, 11, 37],
        "text": "Another angry alien ate an",
        "words": [{
          "boundingBox": [18, 13, 94, 14, 93, 38, 17, 36],
          "text": "Another",
          "confidence": 0.980
        }, {
          "boundingBox": [98, 14, 159, 15, 158, 38, 97, 38],
          "text": "angry",
          "confidence": 0.977
        }, {
          "boundingBox": [163, 15, 212, 15, 212, 38, 163, 38],
          "text": "alien",
          "confidence": 0.978
        }, {
          "boundingBox": [217, 15, 251, 15, 250, 38, 216, 38],
          "text": "ate",
          "confidence": 0.981
        }, {
          "boundingBox": [255, 15, 279, 15, 278, 37, 255, 38],
          "text": "an",
          "confidence": 0.986
        }]
      }, {
        "boundingBox": [11, 49, 311, 48, 311, 72, 11, 73],
        "text": "astounding number of aardvark",
        "words": [{
          "boundingBox": [12, 50, 111, 49, 111, 74, 12, 73],
          "text": "astounding",
          "confidence": 0.962
        }, {
          "boundingBox": [116, 49, 191, 49, 191, 73, 116, 74],
          "text": "number",
          "confidence": 0.981
        }, {
          "boundingBox": [195, 49, 226, 49, 226, 73, 195, 73],
          "text": "of",
          "confidence": 0.986
        }, {
          "boundingBox": [231, 49, 312, 49, 311, 72, 230, 73],
          "text": "aardvark",
          "confidence": 0.690
        }]
      }, {
        "boundingBox": [10, 87, 88, 87, 88, 107, 10, 107],
        "text": "antennae.",
        "words": [{
          "boundingBox": [11, 89, 88, 88, 88, 108, 10, 108],
          "text": "antennae.",
          "confidence": 0.907
        }]
      }]
    }]
  }
}
YutongTie-MSFT commented 3 years ago

We will now proceed to close this thread. If there are further questions regarding this matter, please respond here and @YutongTie-MSFT and we will gladly continue the discussion.

JohnStraumann66 commented 3 years ago

Can you send me the code you used to test this?

YutongTie-MSFT commented 3 years ago

@JohnStraumann66 I didn't use any code, I just use the console to try. https://westus.dev.cognitive.microsoft.com/docs/services/computer-vision-v3-1-ga/operations/5d986960601faab4bf452005

This API is better when you use for heavy-text.