About Text Recognition
Text recognition (OCR: Optical Character Recognition) is a technology that recognizes text in images and converts it into editable text. You can quickly digitize book quotes, handwritten notes, and more without manual typing.
Supported Languages
Reading Forest's text recognition feature supports the following languages:
- Japanese - Kanji, Hiragana, and Katakana
- English - Alphabet and numbers
The text recognition feature supports horizontally written (left-to-right) text only. Vertically written Japanese text is not supported.
On first use, character recognition data (approximately 15MB) is downloaded in the background. Subsequent uses will be much faster.
How to Use Text Recognition
-
Click an Image in Your Notes
Click on an image displayed in the notes area. A context menu will appear -
Select [Read Text from Image]
Choose [Read Text from Image] from the menu. The text recognition window opens with the clicked image loaded in the preview. The image is then automatically analyzed for skew and distortion, and corrections are applied -
Review Auto-Correction
The corrected image is shown in the preview. Use the [Before]/[After] buttons to compare the original and corrected images -
Run Text Recognition
Click the [Read] button. A progress bar shows the processing status. If a book spread is detected, choose from [Left Page]/[Both Pages]/[Right Page] to select the area to read. [Both Pages] processes left and right pages in parallel, with separate progress bars for each -
Review and Edit
The extracted text is displayed. Edit as needed -
Click [OK] to Insert Below the Image
Click [OK] to add the extracted text directly below the clicked image. This action can be undone with [Undo]
Supported Image Formats
The following image formats are supported:
- JPEG (.jpg, .jpeg)
- PNG (.png)
- GIF (.gif)
- WebP (.webp)
- BMP (.bmp)
Recognition Resolution
You can choose from 3 resolution levels when running text recognition. Adjust the balance between speed and accuracy.
Fast
Downscales the image to 1500px before recognition. Quick processing, suitable for rough text extraction
Standard (Default)
Downscales the image to 2500px before recognition. A good balance between speed and accuracy
High
Uses the original resolution for recognition. Better at recognizing small text, but takes longer to process
Your resolution setting is automatically saved and will be used in future sessions.
Auto-Correction
When the text recognition window opens, the image is automatically analyzed and corrected for skew and distortion.
Skew Correction
Detects text line angles and automatically rotates the image to make text horizontal
Page Extraction
Detects the four edges of a photographed page and corrects perspective distortion to extract just the page
Book Spread Support
Automatically detects open book spreads. Choose to read the left page, right page, or both pages. Both Pages mode processes left and right in parallel for faster results
Use the [Before]/[After] buttons in the preview to compare the original and corrected images. If no correction is needed, it is automatically skipped.
Tips for Better Recognition
Clear, Well-lit Images
Use images with good lighting where text is clearly visible
Proper Resolution
Text should be large enough to read clearly in the image
Horizontal Text
Text aligned horizontally yields better recognition results
High Contrast
Higher contrast between text and background improves accuracy
Optimal Shooting Setup
Place documents on a black sheet or board and cover with a glass plate when photographing. This keeps the document flat and reduces shadows, improving recognition accuracy even for small or damaged books
- Angle your lights: Position lights at 45-degree angles from the camera instead of directly behind it to minimize reflections
- Use indirect lighting: Bounce light off the ceiling or white walls for softer, more even illumination
- Adjust camera angle: Shoot from a slight angle rather than directly overhead to avoid reflections. Distortion will be automatically corrected
- Use acrylic instead: Clear acrylic sheets produce less reflection than glass and cost only a few dollars
- Vertically written Japanese text
- Handwritten text (cursive, stylized writing)
- Decorative or artistic fonts
- Blurry images (angled text is auto-corrected)
- Similar colors for text and background
Use Cases
Book Quotes
Photograph a book page and extract the text you want to quote using text recognition. Faster and more accurate than manual typing.
Whiteboards & Blackboards
Capture classroom or meeting notes as searchable, editable text.
Business Cards & Documents
Digitize paper documents and store them in your reading logs for easy organization.
FAQ
Q: Why does it take time on first use?
The first use requires downloading character recognition data (approximately 15MB). This may take a few seconds to a minute. Subsequent uses will be much faster.
Q: Does it work offline?
Yes, once the recognition data has been downloaded, text recognition works offline.
Q: The recognition result is incorrect
Text recognition is not perfect. Always review the extracted text and make manual corrections as needed. See the tips above to improve accuracy.
Q: Can I use it on my smartphone?
Yes, text recognition works in smartphone browsers. You can directly use photos taken with your camera.