How to Use PaddleOCR Only for Text Detection (Skip Recognition)

[kimnamu]

Hi everyone,

I’m happily running the PP-OCRv5 Multilingual model. Thank you very much!
I only need to know where the text is (bounding boxes), not what it says. Below is a concise recipe for running PaddleOCR in detection-only mode and a tip on how to extract bounding-box confidence scores.

[liuhongen1234567]

Hello, if you only need detection, you can refer to the text detection module documentation of PaddleOCR 3.0 at this link: https://paddlepaddle.github.io/PaddleOCR/latest/en/version3.x/module_usage/text_detection.html#3-quick-start.