| 1 | This extension enhances Greenstone3's ability to process with images using
|
|---|
| 2 | the Google Vision API. This includes OCRing of text in documents, OCRing
|
|---|
| 3 | of text in photos, and labelling photos. Presentation of the OCR'd results
|
|---|
| 4 | in enhanced through the Mirador annotation viewer.
|
|---|
| 5 |
|
|---|
| 6 | At build-time, collections can tap into these capabilities using the plugins:
|
|---|
| 7 |
|
|---|
| 8 | GoogleVisionImagePlugin.pm
|
|---|
| 9 | GoogleVisionPagedImagePlugin.pm
|
|---|
| 10 |
|
|---|
| 11 | At runtime
|
|---|
| 12 |
|
|---|
| 13 |
|
|---|
| 14 | #---
|
|---|
| 15 | # Setup
|
|---|
| 16 | #---
|
|---|
| 17 |
|
|---|
| 18 | Run:
|
|---|
| 19 |
|
|---|
| 20 | ./PREPARE-PYTHON3.sh
|
|---|
| 21 |
|
|---|
| 22 | # TLDR
|
|---|
| 23 | cd packages && ./COMPILE-AND-INSTALL.sh && cd ..
|
|---|
| 24 | # See packages/README.txt for more details
|
|---|
| 25 |
|
|---|
| 26 | # TLDR
|
|---|
| 27 | cd src && ./PREPARE.sh && ./COMPILE.sh && ./INSTALL.sh && cd ..
|
|---|
| 28 | # See src/README.txt for more details
|
|---|
| 29 |
|
|---|
| 30 |
|
|---|
| 31 | Review:
|
|---|
| 32 | ant-start-settings.bash.in
|
|---|
| 33 |
|
|---|
| 34 | Copy to:
|
|---|
| 35 | ant-start-settings.bash
|
|---|
| 36 |
|
|---|
| 37 | And edit accordingly. This controls how Greenstone accesses the SimpleAnnotationServer.
|
|---|
| 38 |
|
|---|
| 39 |
|
|---|
| 40 | To enable the Google Vision API, set up a key etc, see:
|
|---|
| 41 |
|
|---|
| 42 | https://cloud.google.com/vision/docs/setup
|
|---|
| 43 |
|
|---|
| 44 |
|
|---|
| 45 | When you create a key, your browsesr downloads a JSON version of the key.
|
|---|
| 46 |
|
|---|
| 47 |
|
|---|
| 48 | #---
|
|---|
| 49 | # Using the Extension: Build-time
|
|---|
| 50 | #---
|
|---|
| 51 |
|
|---|
| 52 | The extension provides the GoogleVisionImagePlugin, for use in collections.
|
|---|
| 53 |
|
|---|
| 54 | Example use of the plugin:
|
|---|
| 55 |
|
|---|
| 56 | <plugin name="GoogleVisionImagePlugin">
|
|---|
| 57 | <option name="-enable_document_ocr"/>
|
|---|
| 58 | <option name="-enable_image_labelling"/>
|
|---|
| 59 | </plugin>
|
|---|
| 60 |
|
|---|
| 61 | The plugin currently has a hard-wired filename for the Google
|
|---|
| 62 | service key. This need's to be in the collection's 'etc'
|
|---|
| 63 | directory as follows:
|
|---|
| 64 |
|
|---|
| 65 | etc/my-google-service-account-key.json
|
|---|
| 66 |
|
|---|
| 67 |
|
|---|
| 68 | For the 'atea-storage' Google Cloud Project, for example, a redacted version of
|
|---|
| 69 | the JSON key file looks like:
|
|---|
| 70 |
|
|---|
| 71 | {
|
|---|
| 72 | "type": "service_account",
|
|---|
| 73 | "project_id": "atea-storage",
|
|---|
| 74 | "private_key_id": "7bc843de4257da515193eaa5b60468fcab5b3948",
|
|---|
| 75 | "private_key": "-----BEGIN PRIVATE KEY-----\n???????.........??????=\n-----END PRIVATE KEY-----\n",
|
|---|
| 76 | "client_email": "[email protected]",
|
|---|
| 77 | "client_id": "123???...????",
|
|---|
| 78 | "auth_uri": "https://accounts.google.com/o/oauth2/auth",
|
|---|
| 79 | "token_uri": "https://oauth2.googleapis.com/token",
|
|---|
| 80 | "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
|
|---|
| 81 | "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/service-account%40atea-storage.iam.gserviceaccount.com"
|
|---|
| 82 | }
|
|---|
| 83 |
|
|---|
| 84 |
|
|---|
| 85 | #---
|
|---|
| 86 | # Using the Extension: Runtime
|
|---|
| 87 | #---
|
|---|
| 88 |
|
|---|
| 89 |
|
|---|
| 90 | Check out the 'intermuse' site and interface
|
|---|
| 91 |
|
|---|
| 92 | Build the intermuse-demo collection
|
|---|
| 93 |
|
|---|
| 94 | Visit the built collection in your browser
|
|---|
