diff --git a/README_en.md b/README_en.md
index a3f95e025812005093836f0b8e8f25eff5ac7c73..b6d6bb1edee2ffbe227e5e1a16a05e506423d68d 100644
--- a/README_en.md
+++ b/README_en.md
@@ -123,7 +123,7 @@ For use of [LSVT](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc
 
 * Note: For the training and evaluation of the above DB model, post-processing parameters box_thresh=0.6 and unclip_ratio=1.5 need to be set. If using different datasets and different models for training, these two parameters can be adjusted for better result.
 
-For the training guide and use of PaddleOCR text detection algorithms, please refer to the document [Text detection model training/evaluation/prediction](./doc/doc_en/detection_en.)
+For the training guide and use of PaddleOCR text detection algorithms, please refer to the document [Text detection model training/evaluation/prediction](./doc/doc_en/detection_en.md)
 
 ## Text recognition algorithm