question

GuptaGaurav01-0474 avatar image
0 Votes"
GuptaGaurav01-0474 asked GuptaGaurav01-0474 answered

Computer Vision API Call not working for Greek Images

Computer Vision API Call not working for Greek Images

I am trying to get Text from images, Text in the image is in the Greek language, I am using Computer Vision 3.2 API

Expected is :Φίλτρο αναζήτ:\n but the result I am getting from analyzeResults API is :ФіАтро аvа(пт:



[1]: /answers/storage/attachments/191450-greektext.png

azure-computer-vision
greektext.png (1.9 KiB)
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

YutongTie-MSFT avatar image
0 Votes"
YutongTie-MSFT answered

Hello @GuptaGaurav01-0474

Print is opposite to HandWritten. The screenshot you shared is print.

Reference: https://wikidiff.com/handwriting/printing

I just got confimation from the pm, we will support Greek preview in Q1 next calendar year so most likely Feb next year, so to answer your whole question:

Chinese, already support
French, already support
Italian,already support
Japanese, already support
Spanish, already support
German, already support
Danish,already support
Norwegian, already support
Finnish, already support
Swedish,already support
Portuguese,already support
Polish, already support
Dutch,already support
Greek, Q1 next calendar year so most likely Feb next year.
Turkish, already support
Korean, already support
Russian already support

If you feel my answer if helpful, please accept it to help the community, thanks.

Regards,
Yutong

5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

YutongTie-MSFT avatar image
0 Votes"
YutongTie-MSFT answered

Hello @GuptaGaurav01-0474

Thanks for reaching out to us, I think you are mentioning the Azure Computer Vision OCR API 3.2 is not working for Greek.

I am sorry for your experience and Greek is not supported now for OCR API 3.2 as below language support page. I just forward your feedback to product group to see the roadmap of Greek lanuage.

https://docs.microsoft.com/en-us/azure/cognitive-services/computer-vision/language-support

Sorry again for the product limitation, I will let you know soon about the roadmap.

Hope this helps!

Regards,
Yutong

-Please kindly accept the answer if you feel it's helpful, thanks a lot.

5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

GuptaGaurav01-0474 avatar image
0 Votes"
GuptaGaurav01-0474 answered YutongTie-MSFT commented

@YoungYang-MSFT : Thanks for your response I am working on OCR for the below language, please let me know the road map for all

Chinese,
French,
Italian,
Japanese,
Spanish,
German,
Danish,
Norwegian,
Finnish,
Swedish,
Portuguese,
Polish,
Dutch,
Greek,
Turkish,
Korean,
Russian

· 1
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Hello @GuptaGaurav01-0474

Are you mentioning print language? Please see below answer, except Greek, all other languages are supported for print language already.

Chinese, already support
French, already support
Italian,already support
Japanese, already support
Spanish, already support
German, already support
Danish,already support
Norwegian, already support
Finnish, already support
Swedish,already support
Portuguese,already support
Polish, already support
Dutch,already support
Greek, I will check with product group
Turkish, already support
Korean, already support
Russian already support

Please kindly accept the answer if you feel helpful, thanks a lot.

Regards,
Yutong

1 Vote 1 ·
GuptaGaurav01-0474 avatar image
0 Votes"
GuptaGaurav01-0474 answered

@YoungYang-MSFT : Thanks a lot for quick reference

Please share documentation of print language classification , I am a beginner in concepts, by print language it means a normal image with text ?

5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

GuptaGaurav01-0474 avatar image
0 Votes"
GuptaGaurav01-0474 answered

Does below are print languages

191926-image.png



image.png (9.1 KiB)
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

GuptaGaurav01-0474 avatar image
0 Votes"
GuptaGaurav01-0474 answered YutongTie-MSFT edited

Hi @YoungYang-MSFT .. Thanks a lot for all your help , I have image with Chinese charaters 192149-chinesechar.png



I tried this image with both Google Cloud vision and Microsoft Cognitive Computer Vision

Somehow i am able to get correct response from Google APi and not from Microsoft API , to me its looks we clear that I a doing something wrong


Google API result

{
"responses": [
{
"textAnnotations": [
{
"locale": "und",
"description": "**输人都門**n",
"boundingPoly": {
"vertices": [
{
"x": 1,
"y": 3
},
{
"x": 70,
"y": 3
},
{
"x": 70,
"y": 62
},
{
"x": 1,
"y": 62
}
]
}
},
{
"description": "输",
"boundingPoly": {
"vertices": [
{
"x": 1,
"y": 11
},
{
"x": 16,
"y": 11
},
{
"x": 16,
"y": 54
},
{
"x": 1,
"y": 54
}
]
}
},
{
"description": "人",
"boundingPoly": {
"vertices": [
{
"x": 18,
"y": 3
},
{
"x": 34,
"y": 3
},
{
"x": 34,
"y": 62
},
{
"x": 18,
"y": 62
}
]
}
},
{
"description": "都",
"boundingPoly": {
"vertices": [
{
"x": 37,
"y": 11
},
{
"x": 52,
"y": 11
},
{
"x": 52,
"y": 53
},
{
"x": 37,
"y": 53
}
]
}
},
{
"description": "門",
"boundingPoly": {
"vertices": [
{
"x": 55,
"y": 13
},
{
"x": 70,
"y": 13
},
{
"x": 70,
"y": 54
},
{
"x": 55,
"y": 54
}
]
}
}
],
"fullTextAnnotation": {
"pages": [
{
"width": 72,
"height": 65,
"blocks": [
{
"boundingBox": {
"vertices": [
{
"x": 1,
"y": 3
},
{
"x": 70,
"y": 3
},
{
"x": 70,
"y": 62
},
{
"x": 1,
"y": 62
}
]
},
"paragraphs": [
{
"boundingBox": {
"vertices": [
{
"x": 1,
"y": 3
},
{
"x": 70,
"y": 3
},
{
"x": 70,
"y": 62
},
{
"x": 1,
"y": 62
}
]
},
"words": [
{
"boundingBox": {
"vertices": [
{
"x": 1,
"y": 11
},
{
"x": 16,
"y": 11
},
{
"x": 16,
"y": 54
},
{
"x": 1,
"y": 54
}
]
},
"symbols": [
{
"boundingBox": {
"vertices": [
{
"x": 1,
"y": 11
},
{
"x": 16,
"y": 11
},
{
"x": 16,
"y": 54
},
{
"x": 1,
"y": 54
}
]
},
"text": "输"
}
]
},
{
"boundingBox": {
"vertices": [
{
"x": 18,
"y": 3
},
{
"x": 34,
"y": 3
},
{
"x": 34,
"y": 62
},
{
"x": 18,
"y": 62
}
]
},
"symbols": [
{
"boundingBox": {
"vertices": [
{
"x": 18,
"y": 3
},
{
"x": 34,
"y": 3
},
{
"x": 34,
"y": 62
},
{
"x": 18,
"y": 62
}
]
},
"text": "人"
}
]
},
{
"boundingBox": {
"vertices": [
{
"x": 37,
"y": 11
},
{
"x": 52,
"y": 11
},
{
"x": 52,
"y": 53
},
{
"x": 37,
"y": 53
}
]
},
"symbols": [
{
"boundingBox": {
"vertices": [
{
"x": 37,
"y": 11
},
{
"x": 52,
"y": 11
},
{
"x": 52,
"y": 53
},
{
"x": 37,
"y": 53
}
]
},
"text": "都"
}
]
},
{
"boundingBox": {
"vertices": [
{
"x": 55,
"y": 13
},
{
"x": 70,
"y": 13
},
{
"x": 70,
"y": 54
},
{
"x": 55,
"y": 54
}
]
},
"symbols": [
{
"property": {
"detectedBreak": {
"type": "LINE_BREAK"
}
},
"boundingBox": {
"vertices": [
{
"x": 55,
"y": 13
},
{
"x": 70,
"y": 13
},
{
"x": 70,
"y": 54
},
{
"x": 55,
"y": 54
}
]
},
"text": "門"
}
]
}
]
}
],
"blockType": "TEXT"
}
]
}
],
"text": "输人都門\n"
}
}
]
}


Cognitive API Result

{
"status": "succeeded",
"createdDateTime": "2022-04-12T05:53:20Z",
"lastUpdatedDateTime": "2022-04-12T05:53:21Z",
"analyzeResult": {
"version": "3.2.0",
"modelVersion": "2022-01-30-preview",
"readResults": [
{
"page": 1,
"angle": -1.7882,
"width": 72,
"height": 65,
"unit": "pixel",
"lines": [
{
"boundingBox": [
0,
9,
17,
9,
18,
54,
2,
55
],
"text": "そ",
"appearance": {
"style": {
"name": "other",
"confidence": 0.915
}
},
"words": [
{
"boundingBox": [
0,
12,
17,
11,
17,
21,
0,
22
],
"text": "そ",
"confidence": 0.019
}
]
}
]
}
]
}
}


chinesechar.png (5.1 KiB)
· 1
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Hello,
You are not inputting right language parameters so the result for both Google and Microsoft are not correct, they did not return Chinese characters.

Please refer to below screenshot to set the right parameter if you are using the API console:
192432-image.png

Regards,
Yutong


0 Votes 0 ·
image.png (14.4 KiB)
image.png (49.2 KiB)
GuptaGaurav01-0474 avatar image
0 Votes"
GuptaGaurav01-0474 answered

Hi @YoungYang-MSFT

Hope you are doing good

I am into new problem , i am not able to get text out of the below image



200078-test1.jpg



test1.jpg (2.3 KiB)
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.