pure browser ocr

Browser-local
image OCR

Recognize text with PaddleOCR and ONNX Runtime Web. Your images are processed locally in this browser and never uploaded.

1Choose an image

No image selected

🖼

Click to choose an image, or drop it here

You can also paste from the clipboard with Ctrl+V · PNG / JPG / WEBP / BMP / GIF

2Recognition results

—

No recognition results yet

Recognition parameters

Detection side limit Binarization threshold Text box threshold Box expansion ratio Minimum text height (px) Small image upscale Recognition width limit Low-confidence threshold

Direction classification (180°) Split long text Filter low-confidence results

Model status

Not loaded

Waiting for initialization

Offline / local models

Models are loaded from the local models/ directory first. If unavailable, online sources are used.

For fully offline operation, use the local model download script and serve this folder through a local HTTP server.

Detection model det.onnx Recognition model rec.onnx Direction model cls.onnx Dictionary ppocr_keys_v1.txt

Browser-localimage OCR

1Choose an image

2Recognition results

Browser-local
image OCR