AI Optical Character Recognition Tool

You are currently viewing the documentation for a pre-release version (2.2.0). To access documentation for other versions, click the "Switch Version" button located in the upper-right corner of the page.

■ If you're unsure about the version of the product you are using, please contact Mech-Mind Technical Support for assistance.

Function Introduction

The AI Optical Character Recognition Tool is a deep-learning visual reading tool for custom character-recognition tasks. It supports a wizard workflow from image capture to inference optimization, and is suitable for difficult scenarios such as engraved metal text, curved-surface text, and complex backgrounds.

The tool includes a pretrained model. After target-region setup, you can run inference directly and then optimize by additional training as needed.

overall workflow

  1. Capture Images: Collect representative images for validation and optimization.

  2. Set Target Region: Set rectangular or annular ROI to focus on text area.

  3. Validate and Optimize: Validate recognition, configure judgment rules, and optionally perform additional training.

Workflow

After opening the tool, click New in model list to create a new model workflow.

Capture Images

  1. Ensure image input is connected.

  2. One image is captured automatically when entering the tool. Click Capture Image for more images.

Include typical variations in capture data:

  • Position and angle variations.

  • Lighting and background variations.

  • Appearance variations such as stains, scratches, and small deformation.

Set Target Region

If 2D Alignment Parameter Group is connected, ROI is transformed synchronously with target poses.

  1. Click Edit to open region settings.

  2. Set single-character size by adjusting the default orange rectangle to match actual character size.

  3. Set target region:

Mode Description

Entire Image as Target Region

ROI covers whole image.

Custom Target Region

Draw rectangle or annulus ROI according to character layout.

  1. Optionally set mask region to exclude glare/shadow/background interference.

  2. Click Save and Use to apply settings.

Validate and Optimize Model

After setting target regions, click Validate.

Set Validation Parameters and Check Results

Parameter Description

Validation Result

Displays OCR output. If judgment is enabled, result is shown as OK/NG based on configured rules.

Time Cost

Inference time per run (ms).

Confidence Threshold

Minimum confidence for character recognition. Results below threshold are considered failed recognition.

Default value: 0.5

Character Content Correction

Constrains character types of leading positions using wildcard template.

Supported wildcards: ? any character, $ letter, % digit, @ symbol, ! uppercase letter, & lowercase letter.

Recognized Character Types

Specifies which character types are considered in downstream processing.

Value list: uppercase letters, lowercase letters, digits, symbols.

Row Delimiter

Delimiter used when multi-line text is recognized.

Enable Judgment

Enables rule-based pass/fail check for recognition output.

Supports:

* Character-count check. * Character-content check (manual template or global-variable baseline string).

Additional Training (Optional)

If validation shows misrecognition, missed characters, or false positives, use additional training:

  1. Click Additional Training.

  2. Add Recognition Content for correction or Exclusion Content for false-positive suppression.

  3. Click Train to retrain with additional data.

  4. Return to validation and iterate until results are acceptable.

  5. Save and apply model: Click Save and Use, then select this model in Model Name for subsequent OCR inference.

Is this page helpful?

You can give a feedback in any of the following ways:

We Value Your Privacy

We use cookies to provide you with the best possible experience on our website. By continuing to use the site, you acknowledge that you agree to the use of cookies. If you decline, a single cookie will be used to ensure you're not tracked or remembered when you visit this website.

  • Preference
  • I accept
  • I don't accept