Skip to content

Latest commit

 

History

History
26 lines (18 loc) · 1.13 KB

README.md

File metadata and controls

26 lines (18 loc) · 1.13 KB

CCPD2YOLO

Process CCPD picture filenames, translate car plate into readable Chinese, draw bounding boxes and make corresponding txt for YOLO.

CCPD paper: Towards End-to-End License Plate Detection and Recognition: A Large Dataset and Baseline

To download CCPD dataset...

Prerequisite:

pip install opencv-python

Just run bbox.py or ccpd2yolotxt.py. Don't forget to replace the default CCPD directory with yours.

Format:

  • CCPD filename format:

    面积比-斜度-左上右下坐标-四个角坐标(右下角开始顺时针)-车牌-亮度-模糊度

  • YOLO txt format:

    0 中心点(归一化宽) 中心点(归一化长) 框宽度(归一化) 框长度(归一化)

Issue:

  • It is not easy for opencv to draw Chinese characters on a picture. You could replace Chinese characters with pinyin if needed. PR is welcomed.