Template

Available in Classic and VPC

In the NAVER Cloud Platform console, under CLOVA OCR > Domain, you can create and manage Template domains, which support a template builder that allows you to directly designate recognition areas, extract values, test results, and send outputs. This section explains how to create and use a Template-type domain. For instructions on viewing, deleting, and copying domains, as well as detailed steps for integrating with API Gateway, see Domain.

Create a domain

To create a Template-type domain:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Create domain] button, then select [General/Template] button.
4. When the domain creation popup appears, enter the required information.
   • Domain name: Enter 1 to 50 characters using Korean, English (uppercase/lowercase), Chinese, Japanese, numbers, or full-width/half-width special characters ( & 。_ - × 々 ). Click [Check] button to verify duplication.
   • Domain code: Enter 1–50 characters using English letters and special characters (.?&。_-×々). Spaces are not allowed. Click [Check] button to verify duplication.
   • Supported languages: Select one or more languages supported by the domain.
   • Service type: Select Template
   • Recognition model: Select the recognition model to apply to the domain.
   • Service plan: Select the service plan to apply to the domain.
5. Click [Create domain] button.

Run the template builder

After creating a Template domain, you can launch the template builder. The template builder is a web-based console used to create and test templates for extracting text from images or documents. It provides a variety of features required for Template OCR. The template builder consists of the following feature menus:

• Create template: Create a primary template and manage representative sample lists for templates to be analyzed.
• Test & Analyze: Test templates/models with designated recognition areas, review results, and review OCR statistics by service plan.
• Components: Register specialized words and synonyms in the terminology dictionary, apply them to templates, and manage terminology entries.
• Settings: Review domain information, configure external integration, set up validation integration, and manage permissions.
• Deployment management: Manage beta deployments and service deployments for templates.

Create a template

In the Create template menu, you can create a primary template and manage the representative sample list for each template used for analysis.

Template list

To create a primary template and manage sample files:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. In the Template builder, click the **Template list ** menu.
5. Click [Create template] button.
6. In the Template basic information section, enter the template name and click [OK] button.
   • Template name: Enter 1-30 characters using Korean, English (uppercase/lowercase), Chinese, Japanese, numbers, or full-width/half-width special characters (? & 。 _ - × 々).
   • Recognition model: Displays the recognition model applied to the selected
7. Click [OK] button.
8. In Representative sample settings, upload a sample image for recognition performance.
   • Image file upload requirements:
     • File size: 20 MB or less
     • File format: jpg, png, pdf, tiff
     • File resolution: At least 150 dpi for A4; minimum 10 px, recommended up to 1960 px on the longer side
     Note

     Images submitted through API should be high quality, at least 150 dpi, with minimal distortion.

9. After the sample image is uploaded, specify the document title area.
   
   • Specify title area: After uploading a sample image, drag the selection rectangle to mark the title area.
     • Clicking an existing area activates the selection rectangle. Click to remove the area.
     • Clicking an existing area activates the rectangle; drag the dotted line to resize it. Click to save the updated area.
10. In Representative sample name, enter the name used to classify the document template. Use 1–30 characters with Korean, English (uppercase/lowercase), Chinese, Japanese, numbers, or full-width/half-width special characters ( ? & 。 _ - × 々 ), then click [OK] button.
    
    • When you click [OK] button, a notification appears for saving and synonym registration.
    • After clicking [OK] button, you can modify the representative sample name later by clicking [Edit].
    • [Manage synonyms] button: Click to register words that may be recognized as similar to the representative sample name to improve classification performance. If you want to enter multiple synonyms, separate them with commas.
      
      • For example, Local tax bill, local tax, local tax notice
      • The representative sample name in Synonym Management cannot be modified. You can only register or edit synonyms.
11. In Field Definition, specify and configure each field you want to extract.
    • When clicking Add field
      • Define recognition area: When a new field selection box appears, adjust the size of the box to set the area.
        
        • Clicking an existing area activates the selection rectangle. Click to remove the area.
        • Clicking an existing area activates the rectangle; drag the dotted line to resize it. Click to save the updated area.
        • When you click , the field name (original name_#) and its configuration values are duplicated, and a new field box is created below the existing field.
        • Click [Delete] button in the field box to delete the field.
    • When selecting Multi Box (Premium only)
      
      • If the input fields are divided, such as bank account numbers or mobile phone numbers, you can use Multi Box to recognize and extract characters inside each individual box.
      • If you set the field area outside the Multi Box boundary, only the characters inside the intended boxes are recognized (the Multi Box boundary itself is not recognized).
      • Define recognition area: When a new field selection box appears, adjust the size of the box to set the area.
        • Clicking an existing area activates the selection rectangle. Click to remove the area.
        • Clicking an existing area activates the rectangle; drag the dotted line to resize it. Click to save the updated area.
        • When you click , the field name (original name_#) and its configuration values are duplicated, and a new field box is created below the existing field.
        • Click [Delete] button in the field box to delete the field.
    • When you click Checkbox (Premium only)
      
      • If an irregular character such as V, X, O, or · is detected inside a checkbox, such as a selection box or radio button, the result returns True. If no such character is detected, it returns False.
      • OCR can recognize check marks across various types of documents, including surveys and product application forms.
      • When configuring a checkbox, make sure the field is set so that the input box boundary is clearly visible, and ensure that only the checkbox itself is recognized within the designated area, without any additional text or shapes. The Multi Box boundary and the field area box should be aligned as closely as possible.
      • If an irregular character is located outside the checkbox area, if the character is erased or faint, if the same character is overwritten during correction, or if irregular characters or guide marks are already printed inside the checkbox for user marking, accurate recognition may not be possible.
      • Define recognition area: When a new field selection box appears, adjust the size of the box to set the area.
        • Clicking an existing area activates the selection rectangle. Click to remove the area.
        • Clicking an existing area activates the rectangle; drag the dotted line to resize it. Click to save the updated area.
        • When you click , the field name (original name_#) and its configuration values are duplicated, and a new field box is created below the existing field.
        • Click [Delete] button in the field box to delete the field.
    • Unrecognized area: If necessary, you can mark specific parts of a field area as unrecognized so that OCR does not read characters or handwritten content in that area.
      • To designate an unrecognized area within an existing field: Click Add field, then specify the area inside the existing field that should not be recognized.
        • Text inside an area set as unrecognized will not be extracted.
      • Using Masking (supported only for standard fields with Value type = All):
        • Not supported for Multi Box, Check Box, or Value type = Numeric.
        • If guide text (background text) is printed inside the field, such as instructions for writing, unrecognizing background text or watermarks is not supported.
12. Enter a name for the added field and configure the relevant options.
    • Enter a field name: Enter the field name using Korean, English (uppercase/lowercase), Chinese, Japanese, numbers, and full-width/half-width special characters ( ? & 。 _ - × 々 ).
      • You can freely modify the field name at any time.
    • Field settings: Click
      • Value type and term settings: Applies when selecting a standard field or Multi Box. Value type:
        • Value type: When you select All, both text and numbers can be recognized. When you select Numeric, only numbers are recognized. By specifying the expected value type in advance, you can ensure that only the intended values are extracted from the field.
        • Term registration: You can register a terminology dictionary containing domain-specific terms and synonyms. When a dictionary is applied, only the representative terms are recognized. If a synonym is detected, it is automatically converted to the representative term. Recognition of any word not included in the dictionary is treated as a failure.
      • Checkbox output conversion: Applies when selecting a Check Box field.
        • You can convert the checkbox's return values (True (Y) and False (N)) into specific characters for output.
          
13. If necessary, configure field alignment or combine field recognition results.
    • Field alignment: Hold Shift key, drag to select the fields to align, then click Align left or Align top in the field alignment options.
      
    • Combining extraction results: To combine OCR extraction values from individual fields and output them as a single new value, click [Combine extraction results] button.
      • You can extract a processed OCR result according to the fields selected for combination and the specified output order.
      • If you enter fixed text, the output appears as “each field's output value + fixed text.”
14. Click [Save] button.
15. Manage your templates as needed in the template list.
    
    • Change representative sample image: Click [Edit] button in the representative sample list.
      • When you change the representative sample image, all previously saved content is deleted.
    • Delete a template: Select the template you want to delete in the representative sample list, then click [Delete] button.
    • Template validation: If you want to validate each template by connecting your own validation server, click [Register] button in the validation section.

Sample list

You can view and download representative image samples registered for each template. To view and download sample images:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click the Sample list menu under Template creation.
5. When the sample list appears, review the information and download image samples as needed.
   • View uploaded representative sample image: Click an individual sample number.
   • Download uploaded representative sample image: Click in Download

Test & Analysis

In Test & Analysis, you can upload an image to test text extraction and download the extraction results. You can also view statistical data such as daily API call volume by service plan, external validation results by template, and recognition result metrics.

Test

To upload an image for text extraction and test or download the extraction results:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click the Test menu under Test & Analysis.
5. In the test conditions, select Beta or Template extraction.
   • Bata: Tests template auto-classification and extraction areas in an environment identical to production. After Beta deployment, you can run Beta tests and then deploy to Service.
   • Template extraction: Tests templates and models with configured extraction areas.
6. Upload the test data under Test data.
7. Review the results under Test results.
   
   • View test results: Results are displayed in a table.
   • Download test results: Select csv or json, then click [Download]

Metrics

To view daily API call volume by service plan, external validation results by template, and recognition result statistics:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click the Metrics menu under Test & Analysis.
5. Set the date range and the template.
   • Date range: Up to 90 days.
6. Review the results.
   • API call volume: Number of API calls made during the selected period for the active service plan.
   • Validation results: External validation results by template and the number of validation attempts for the domain. When you hover over the validation chart, VALID, UNCHECKED, ERROR, and INVALID counts are displayed.
   • Template recognition failures: If a template does not match and no recognition result is returned, it is counted as a template recognition failure. The number of failures during the selected period is displayed.
7. To download the query results as an xls file, click [Download integrated metrics] button.

Terminology Management

In Terminology Management, you can register domain-specific terms and synonyms into a terminology dictionary and apply them to templates. You can configure a field to recognize only the terms contained in the dictionary, and if any registered synonym is recognized, it is replaced with the representative term. To configure a terminology dictionary:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click the Terminology dictionary menu under Terminology Management.
5. Click [Create terminology dictionary] button.
6. When the dictionary creation pop-up appears, enter the required information.
   
   • Dictionary name: Enter a unique name that does not duplicate existing dictionary names.
   • Representative term: Enter the representative term used to define the dictionary.
   • Synonyms: Enter synonyms or similar words for the representative term. When adding multiple synonyms, separate them with commas. For example, Local tax bill, local tax, local tax notice
7. Click [OK] button.

Settings

In Settings, you can view domain information or configure options such as external integration and permission management. To view or configure settings:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click [Settings] button.
   
5. Click the desired tab to view or configure the relevant settings.
   • [Domain information]: View domain and template information.
   • When [API Gateway integration] button is clicked: Configure API Gateway integration.
     • Refer to Domain for detailed instructions.
   • [External validation server]: Click [Create] button to register and manage the URL where OCR results will be transmitted.
     
     • Enter the name in the format ${Result transmission name}
     • Enter a URL starting with https:// in the URL field.
   • [Permission management] tab: Adds new administrators and grant permissions.
     • Click [Generate URL] button to create an Invite URL, which you can use to grant permissions.
     • Click [Delete] button to delete the Invite URL.

Deployment management

Beta deployment is the final save function that applies changes to the model. It allows the system to learn the samples, reading areas, and terminology dictionary that you configured in the template. After performing a beta deployment, you can run beta tests and then proceed with service deployment.

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click [Deployment management] button.
   
5. When the template list appears, review the latest status of each template under the current deployment state.
6. Select the template you want to deploy and click [Beta deploy] button.
   • After completing a beta deployment, you can run beta tests from the Test menu.
   • If you change anything in the Test menu after testing, apply it by performing a new beta deployment.
7. After beta testing is complete, click [Service deploy] button to deploy the latest beta-deployed template to production.

Integrate with API Gateway

To configure the Invoke URL for Template OCR, integrate with API Gateway as follows:

1. In the NAVER Cloud Platform console, navigate to > Services > AI Services > CLOVA OCR.
2. Click Domain.
3. Click [Template builder] button for the Template domain.
4. Click the [API Gateway integration] tab, then click [Integrate] button.
5. When the API Gateway integration popup appears, configure the required information.
   • For detailed instructions, see Domain > API Gateway integration.

Call the API

When the API Gateway integration popup appears, enter the required information. For detailed instructions, see Domain > API Gateway integration. For more information about calling the API, see:

• CLOVA OCR API Guide
• Template OCR API call examples