About OCR Image to Text
Recognises printed, handwritten, horizontal and vertical text in images with the browser-local PaddleOCR PP-OCRv6 model and WebAssembly inference. Supports Simplified Chinese, Traditional Chinese, English, Japanese and Latin-script languages. Supported image formats: png, jpg, bmp, pbm.
Input: binary bytes (text input uses UTF-8; a file retains its original bytes). Output: text.
Format reference: Optical Character Recognition background and specification
Example
Use “Try example” to load this public sample and its settings, then run it.
Settings used in this example
| Parameter | Example value | Input and choices |
|---|---|---|
| Show confidence | false | On or off |
Result
The example demonstrates the configured operation. Generated values, timestamps or formatting may vary between runs.
Parameters
| Parameter | Default value | Input and choices |
|---|---|---|
| Show confidence | true | On or off |
Before interpreting the result
Open a local file for binary media. Supported formats depend on the operation and decoder. Large images are bounded by pixel and memory limits; previews may differ from the original downloaded bytes.
How to use this page
- Paste your input or open a local file.
- Adjust tool options if needed, then choose Run tool.
- Copy the result or download the original output bytes.
Inputs and results are not added to the public catalogue or saved in the URL. Processing happens in your browser; HTTP Request and DNS over HTTPS send the requests you explicitly run. Files up to 128 MiB can be opened; decompression and image tools may have smaller processing limits. Large results have a shortened preview. Text and HTML downloads use UTF-8; binary downloads preserve output bytes. Pasted text follows browser newline handling; open a file when exact input bytes matter.