Google Cloud Vision image analysis
"Google Cloud Vision" is a cloud service that Google offers for image analyses. censhare integrates the Google Cloud Vision service to analyze images stored in your system.
Prerequisites
The Google Cloud Vision functionality is available for Image assets with the MIME types JPEG, PNG, or GIF.
To run automatic analyses, the automatic server action for Google Cloud Vision must be configured.
To start an analysis manually, the manual server action must be configured.
Introduction
Google Cloud Vision is a cloud service for image analysis. censhare uses this service to add information as properties to images.
censhare offers the following functionality from the Google Cloud Vision API:
Assign keywords
Identify the main colors of an image
Find the image or similar images on the web
Find web pages that contain the image
Identify locations shown in the image
Identify brands related to the image
Detect text shown in the image
Check for inadequate content (safe search)
Depending on your system configuration, not all functions are available. If you need a function to be activated, contact your system administrator.
For more information about Google Cloud Vision, see https://cloud.google.com/vision/.
censhare analyzes the following asset types:
"Image" assets with JPEG MIME type
"Image" assets with GIF MIME type
"Image" assets with PNG MIME type (PNG 8 bit and PNG 24 bit)
censhare shows the results of an analysis on the "Analysis" tab on an Image asset page. Several widgets present different results. On the “Analysis” tab, you also find the widget to start a new analysis.
Note: Be aware that censhare is dependent on the data provided by Google. Only results provided by Google after an analysis can also be shown in censhare.
The Google Cloud Vision results in detail
The following table shows which Google Cloud Vision results censhare uses and how they are mapped in censhare:
Google category | Google sub-category | censhare result | censhare sub-result |
Web | Web Entities | Keywords | - |
Web | Pages with Matched Images | Matching images | Full matching image page (URL) |
Web | Fully Matched Images | Matching images | Full matching image |
Web | Partially Matched Images | Matching images | Partial matching image (URL) |
Properties | Dominant Colors | Content Colors | - |
Safe Search | - | Safe Search | - |
Landmarks | - | Locations | - |
Logos | - | Brands | - |
Document | - | Texts | - |
For more information about the different categories delivered by Google and their processing in censhare, see Concept of Google Cloud Vision analysis.
Assign keywords
The "Keywords" widget on the Analysis tab shows all keywords that have been identified and assigned to an asset. The relevance score of a keyword is shown next to the name of the keyword. Double-click a keyword entry to open the corresponding Keyword asset.
On the Keyword asset page, the "Keyword of" widget on the Overview tab shows all assets to which this keyword is assigned. Double-click a list entry to open one of these assets.
If you want to see a complete list of keywords that were created by the Google Cloud Vision integration, search for the asset type = "Module/Keyword" and the "External source provider = "censhare:external-source.google". Use the Expert search for this query.
Main colors of an image
The "Content colors" widget on the Analysis tab shows which colors from the 16-color palette in censhare were detected. Besides the color, censhare shows a percentage. The number corresponds to the calculated coverage of that color in the image.
Note: As of the mapping from the Google colors to the 16-color palette, the colors and values shown in the “Content colors” widget are only approximate values.
Related web pages
Google Cloud Vision searches in three web-related categories:
Images that have a full match with the analyzed image
Images that have a partial match with the analyzed image
Web pages that contain a full match of the analyzed image
The "Matching images" widget shows the URLs for all three categories. The headline for each category also shows the number of results for this category. Each URL entry is linked to the Web. Click an entry to open the URL in another page in your Web Browser.
Locations
The "Locations" widget on the Analysis tab shows all landmarks that were detected. The relevance score of a location is shown next to the name of the location. Double-click a location entry to open the corresponding Location asset.
On the Location asset page, the "Location of" widget on the Overview tab shows all assets to which this location is assigned. Double-click a list entry to open one of these assets.
For a complete list of locations created by the Google Cloud Vision integration, search for the asset type = "Location" and the "External source provider = "censhare:external-source.google". Use the Expert search for this query.
Brands identified
The "Brands" widget on the Analysis tab shows all logos that were detected. Google calculates a relevance score for the brand logo. The relevance score is shown next to the name of the Brand asset. Double-click a location entry to open the corresponding Brand asset.
For a complete list of brands created by the Google Cloud Vision integration, search for the asset type = "Brand" and the "External source provider = "censhare:external-source.google". Use the Expert search for this query.
Text detected
Google Cloud Vision uses Optical Characters Recognition (OCR) to find text in the analyzed image. If Google detects text, censhare stores the result as plain text in a storage item attached to the image.
Even if the text is visible in the image, Google cannot always recognize it. For instance, if the text is only partially visible on a T-Shirt, no text result can be shown in censhare.
The "Texts" widget shows any detected text.
You can search for text that Google has detected. In the Detailed search or the Expert search, use the "Content (full text)" field.
To download the OCR text, use the "Download files" action in the page action menu. In the Dialog, expand the "Show more options" and select the "Text preview" entry.
Safe search results
The "Safe Search" feature of Google Cloud Vision checks an image for inadequate content. The inadequate content includes five categories:
Category | Remark |
Adults only | Proximity that the image is for adults only |
Spoof | Proximity that the image has been modified from the original to make it funny or offensive |
Medical | Proximity that the image is related to medical topics |
Violence | Proximity that the image is related to violent topics |
Racy | Proximity that the image is related to racy topics |
The "Safe search" widget shows the results.
How to execute the Analysis
How to start Google Cloud Vision analysis:
From the “Analysis” tab of the Image asset page
From the Page actions menu of the Image asset page
From the Asset actions menu of an Image asset entry on an asset list page
Directly from Analysis tab
Open the image that you want to analyze in censhare Web.
Go to the "Status" widget on the "Analysis" tab.
Click ANALYZE VIA GOOGLE SERVICES.
The Status widget shows the time and date when the last analysis was done. When a new analysis is finished, censhare updates the date and time entries. If no analysis was executed yet, the widget shows the message "Not analyzed yet".
From the page action menu
Open the image, you want to analyze, in censhare Web.
Click the page action menu on tab bar.
Select "Google Vision" in the sub-menu "Others".
From an asset list page
Search for the image that you want to analyze.
Click on the asset action menu of the desired asset list entry.
Select "Google Vision" in the sub-menu "Others".
Note: You can only start one analysis at a time. If you select more than one asset, the "Google Vision" entry is not available. If you want to start more than one analysis at one, go to the censhare Client.
Result
The "Google Cloud Vision" server action is executed for the selected Image asset and the results are stored in censhare. The results of the analyzes are shown in the “Analysis” tab of the selected Image asset.
Next steps
No additional steps are necessary.