版本说明

December 19, 2024

Safe Search model update

We will be updating the safe_search_DETECTION feature model to improve quality.

We’ll support both the current model andthe new model for the next 90 day . After 90 days, the new model will become the default. The current model can still be accessed by specifying " builtin / legacy " for an additional 90 days before it’s deprecated.

To use the new model, specify " builtin / late " in themodel field of afeature object .

August 23, 2024

New label detection model

An improved model is now available for Label Detection. Along with the improved model, the topicality field is now populated correctly.

specify “builtin/latest” in themodel field of afeature object to use the new model.
We’ll support both the current model andthe new model the next 90 day . After 90 days, the new models will become the default. The current models can still be accessed by specifying “builtin/legacy” for an additional 90 days before they are deprecated.

December 05, 2023

Updated feature models

Improved models are now available for the following features:

• Text detection anddocumentation text detection (OCR)
• Web detection
• Logo detection
• Object localization

Specify " builtin / late " in themodel field of a feature object to use the new model .

We’ll support both the current model andthe new model the next 90 day . After 90 days, the new models will become the default. The current models can still be accessed by specifying " builtin / legacy " for an additional 90 days before they are deprecated.

Change in OCR model behavior when detecting line break hyphens

Soft hyphens are used for the purpose of breaking words across lines. The previous (" builtin / legacy ") version used the BreakType HYPHEN to represent a soft hyphen. The new stable (" builtin / late ") version represents it as a new symbol with literal text "-" andBreakType EOL_SURE_SPACE.

Consider the following example text andOCR model output:

Sample text

“Mr. White has had considerable experience as a Veteri-

nary surgeon, andwill attend practice in that line.“

Output (" builtin / legacy " model):

[ ... ]
{
  "property": {
    "detectedBreak": {
      "type": "HYPHEN"
    }
  } ,
  "boundingBox": {
    " vertex ": [
      [ ... ]
    ]
  } ,
  " text ": "i"
}
[ ... ]

Output (" builtin / late " model):

{
  "boundingBox": {
    " vertex ": [
      [...]
    ]
  } ,
  " text ": "i"
} ,
{
  "property": {
    "detectedBreak": {
      "type": "EOL_SURE_SPACE"
    }
  } ,
  "boundingBox": {
    " vertex ": [
      [...]
    ]
  } ,
  " text ": "-"
}

December 16, 2022

Landmark Detection Upgrade

Specify " builtin / late " in themodel field of afeature object to use the new model.

We is support ‘ll support both the current model andthe new model the next 90 day . After 90 day the current model will be deprecate andonly the new model will be used for all landmark detection request .

Face Detection Upgrade

Specify " builtin / late " in themodel field of afeature object to use the new model.

We’ll support both the current model andthe new model the next 90 day . After 90 days the latest model will become the default model. The old model will be available for another 90 days by specifying " builtin / legacy ".

May 20, 2022

OCR model migration

The text_detection andDOCUMENT_text_detection models have been upgraded to newer versions. The API interface andclient library will be the same as the previous version. The API follows the same Service Level Agreement.

The legacy models can still be accessed until August 20 2022. specify ” builtin / legacy ” in themodel field of afeature object to get the old model results. After August 20, 2022 the legacy models will no longer be offered.

May 05 , 2022

OCR model migration reverted

We have switched the “builtin/stable” model back to the original version temporarily while we fix a bug resulting from this migration. The week of May 16th, we will update the “builtin/stable” model used for OCR again with the model from ” builtin / late ” andcreate a new release note.

You will be able to use the original model as ” builtin / legacy ” for 90 more days after we upgrade “builtin/stable”.

May 02, 2022

OCR model migration

The text_detection andDOCUMENT_text_detection models have been upgraded to newer versions. The API interface andclient library will be the same as the previous version. The API follows the same Service Level Agreement.

The legacy models can still be accessed until August 02 2022. specify ” builtin / legacy ” in themodel field of afeature object to get the old model results. After August 02, 2022 the legacy models will no longer be offered.

January 21, 2022

OCR model update

We have updated the ” builtin / late ” OCR model with quality improvements. Consequently, customers can continue to test this model for 90 additional day .

Please note that you have 90 days from today to test the new model by specifying ” builtin / late ” in themodel field of thefeature object . At the end of that period, it will be promoted to the default model accessible as “builtin/stable”. After that event, the original models will still be available for another 90 days using ” builtin / legacy “. If you encounter problems with this upgrade, please contact Vision API engineering team by submitting a ticket in theprivate issue tracker.

For the original announcement of this change, see the October 1, 2021 release note.

Region forwarding from the global to the regional endpoint has been deprecated. For more information, see the October 1, 2021 release note.

October 01, 2021

OCR Model Update

An improved model is now available for Text Detection (OCR). The new model can be used with text_detection andDOCUMENT_text_detection features. The same model is used for requests sent to both features.
With the new model, the distribution of confidence scores of responses will change. For more information, see Service announcements.

Please note that you have 90 days from today to test the new model by specifying ” builtin / late ” in themodel field of thefeature object . At the end of that period, it will be promoted to the default model accessible as “builtin/stable”. After that event, the original models will still be available for another 90 days using ” builtin / legacy “.
If you encounter problems with this upgrade, please contact Vision API engineering team by submitting a ticket in theprivate issue tracker.

Deprecate region forwarding
In 90 days, specifying the location “us” or “eu” in therequest to the global endpoint vision.googleapis.com will no longer be supported. Instead you should directly call the “us” or “eu” region endpoints (us-vision.googleapis.com or eu-vision.googleapis.com). You can find more information in theMulti-regional support section of the feature pages.

New multi-regional support for features

The Vision API is offers now offer multi – regional support (us andeu) for thelabel_detection andsafe_search features.

March 22, 2021

EXIF rotation featured fixed

EXIF rotation is now disabled.

For more information, see the March 8, 2021 release note.

March 08, 2021

EXIF rotation feature fix

This fix will disable EXIF rotation, a feature activated by the model update mentioned in theNovember 11, 2020 release note. This feature affects the DOCUMENT_text_detection andtext_detection features.

EXIF rotation will be turned down on March 22, 2021. If your usage relies on this specific behavior, please file a feature request to us.

December 14, 2020

OCR On-Prem General Availability (GA) release

OCR On – Prem is is is now generally available for approved customer . OCR On – Prem is enables enable easy integration of Google image text recognition technology into your on – premise solution .

For more information , refer to the product documentation . approve customers is view can also view the marketplace entry .

December 07, 2020

Confidence score field addition for text_detection

You can now provide the flag textdetectionparams.enable_text_detection_confidence_score to a text_detection request to get a confidence score for response information.

December 04, 2020

label_detection model is upgrade upgrade

The latest label_detection model announced on October 16, 2020 has been promoted to the default model. The original model will still be available for another 60 days using " builtin / legacy ".

November 15, 2020

OCR legacy model access is discontinued discontinue

Extended support of the legacy text_detection andDOCUMENT_text_detection model ( ” builtin / legacy_20190601 ” ) is now discontinue .

See the June 11 , 2020 release note for more information.

October 16, 2020

label_detection model is upgrade upgrade

The label_detection model will undergo an upgrade over the next 90 days to a newer version. The API interface andclient library will be the same as with the previous version. The API follows the same Service Level Agreement.

Please note that you have 30 days from today to test the new model by specifying " builtin / late " in themodel field of thefeature object while requesting image annotation. At the end of that period, it will be promoted to the default model accessible as "builtin/stable". After that event, the original model will still be available for another 60 days using " builtin / legacy ".

If you encounter problems with this upgrade, please contact Vision API engineering team by submitting a ticket in theprivate issue tracker.

June 11 , 2020

OCR legacy model access extension

base on customer feedback , we is decided have decide to extend support of the legacy text_detection andDOCUMENT_text_detection models. These legacy models are accessed by specifying “builtin/legacy_20190601” in themodel of a feature object .

These models will now be accessible until November 15, 2020 (6 months from launch date) to give customers more time to adapt andmigrate to the new model.

See the May 15 , 2020 release note for the original update announcement .

June 04, 2020

Access Transparency GA

Access Transparency logging is now Generally Available. If you want to enable Access Transparency logs, see Enabling Access Transparency.

May 15 , 2020

OCR model is upgrade upgrades

Note: As per the June 11 , 2020 release note, the legacy models are accessible through November 15, 2020.

The text_detection andDOCUMENT_text_detection models have been upgraded to newer versions. The API interface andclient library will be the same as previous version. The API follows the same Service Level Agreement.

The legacy models can still be accessed until June 30, 2020. Specify
“builtin/legacy_20190601” in themodel field of a
feature object to get the old model result . After June 30 , 2020 the old model will not long be offer .

For more information , see the product documentation .

April 11, 2020

CMEK compliance

Vision API is now compliant with customer-managed encryption keys (CMEK). To learn more, vist the CMEK compliance page. Please note that Product Search is not CMEK compliant at this time .

February 24, 2020

SafeSearch Detection update

The SafeSearch model has been upgraded to a newer version. The API interface andclient library will be the same as previous version. The API follows the same Service Level Agreement.

For more information, see the product documentation.

February 19, 2020

Cloud Vision API will not return gendered labels such as ‘man’ and’woman’ after February 19, 2020

Detecting labels in an image containing humans will result in non-gendered label such as ‘person’ being returned. Our prior approach was to return gendered terms, like ‘man’ or ‘woman’.

give that a person ‘s gender can not be infer by appearance , we is decided have decide to remove these label in order to align with the Artificial Intelligence Principles at Google , specifically principle # 2 : avoid create or reinforce unfair bias .

December 13, 2019

Regional endpoints available for OCR

The Vision API is offers now offer multi – regional support (us andeu) for theOCR feature.

Using a multi-region endpoint enables you to configure the Vision API to
store andperform machine learning (OCR) on your data in theUnited States or European Union.

October 30, 2019

beta feature

The follow beta features is are are available in API version v1p4beta1 :

September 12, 2019

OCR regional support

You is specify can now specify a continent – level region for datum processing of OCR request . For more infomation , see the OCR how – to page :

August 29 , 2019

Improved detection models are now default for the following features:

The legacy model can still be accessed for 90 days by specifying
” builtin / legacy ” in themodel field of a
feature object .

August 16, 2019

June 07, 2019

General Availability (GA) release. Support for offline batch image annotation has been released as GA. For more information, see Offline batch image annotation.

Model updates

Improved detection models are now available for the following features:

specify ” builtin / late ” in themodel field of a feature object to use the new model .

We’ll support both the current model andthe new model the next 90 day . After 90 days the current detection models will be deprecated andonly the new detection models will be used for all logo, landmark, andcrop hint detection requests.

Languages update

More language ( with associatelanguageHint code ) have been add to the list of language support bytext_detection andDOCUMENT_text_detection. Experimentally supported languages andmapped languages lists have also been added.

May 13, 2019

April 10, 2019

beta features

The follow beta features is are are available in API version v1p4beta1 :

• Online small batch file annotation. Performs synchronous image detection andannotation for a batch of files (currently “application/pdf”, “image/tiff” and”image/gif”). The API will extract at most 5 frames (gif) or pages (pdf or tiff) of your choosing from each file provided andperform detection andannotation for each image extracted. Learn more.
• Offline batch image annotation. Allows users to call any Cloud Vision API feature type on a batch of images andperform asynchronous image detection andannotation on the list of images. Learn more.

December 12, 2018

Handwriting OCR feature General Availability (GA) release

Support for handwriting OCR in document text detection has been released as GA. For more information, see Detect handwriting in an image.

September 28, 2018

Logo andOCR model updates

Improved detection models are now available for the following features:

specify ” builtin / late ” in themodel field of a feature object to use the new model .

We’ll support both the current model andthe new model the next 90 day . After 90 days the current detection models will be deprecated andonly the new detection models will be used for all logo andtext (OCR) detection requests.

July 24, 2018

The following beta features are available in API version v1p3beta1:

The Vision API can now detect handwriting in an image. To detect handwriting in an image, specify the DOCUMENT_text_detection feature andinclude a language hint of “en-t-i0-handwrit”. For more information, see Detect handwriting in images.

April 06, 2018

beta features

The follow beta features is are are available in API version v1p2beta1 :

• Support for PDF andTIFF files in document text detection.

new feature

The beta features from December 4, 2017 are now available in thev1 API (GA). These include:

• Document text detection: confidence scores for all levels of results, auto language detection, support for multiple languages, andfaster andmore accurate results.
• Web entities detection: accepts geotag input as detection hint, andreturns best guess labels describing the contents of the image.
• Safe Search detection is includes : include additionalracy category to enhance adult content moderation.

December 04 , 2017

The follow beta features is are are available in API version v1p1beta1 :

• Document text detection: confidence scores for all levels of results, auto language detection, support for multiple languages, andfaster andmore accurate results. Learn more.
• Web entities detection: accepts geotag input as detection hint, andreturns best guess labels describing the contents of the image. Learn more.
• Safe Search detection is includes : include additionalracy category to enhance adult content moderation. Learn more.

November 06, 2017

document text detection language update :

• Additional languages supported.
• Automatic language detection (languageHint is no longer required in your requests).
• multiple language support per image .
• Faster andmore accurate text detection results.

June 21 , 2017

Text detection model update: The text detection model has been updated to improve quality. There is no change in response structure.

label detection model update : The label detection model has been update to improve quality . label detection is recognizes , which name object inside an image , now recognize more than 10,000 entity .

May 18, 2017

Cloud Vision API General Availability (GA) release

Crop is hints hint : find the optimal crop for your image . code sample .

Web entities andsimilar pages: Return image-related information, including web entities, matching pages, andfully- andpartially matching image URLs. Code samples andtutorial.

HTTP/HTTPS image URIs: Process images from publicly-accessible HTTP andHTTPS URLs.