使用paddleocr进行文字识别_paddleocr实现文字识别

配置环境

pip install paddleocr
pip install paddlepaddle
1
2

输入中文

PIL模块使用Windows中"C:\Windows\Fonts"目录下自带的中文包。

代码

from paddleocr import PaddleOCR, draw_ocr
import cv2
from PIL import ImageFont,ImageDraw,Image
import numpy as np

def paddle_ocr(img_path):
    orc = PaddleOCR(use_angle_cls=True, use_gpu=True)
    font1=ImageFont.truetype("./STSONG.TTF",15)
    image_cv = cv2.imread(img_path)
    img = np.zeros_like(image_cv) + 255
    img_draw= Image.fromarray(img)
    draw = ImageDraw.Draw(img_draw)
    result = orc.ocr(img_path, cls=True)


    for i  in result:
        for k in i:
            left_button, right_button, right_down, left_down = k[:-1][0]
            point_1 = [int(x) for x in left_button]
            point_2 = [int(x) for x in right_down]
            # draw_point = [point_1[0], point_1[1]-10]
            values, confidence = k[-1]
            print("k:",k)
            # print(index,"00000")
            print("index:", left_button,right_button,right_down, left_down)
            print("valves:", values)
            print("confidence:", confidence)
            cv2.rectangle(image_cv, point_1, point_2, (0, 0, 255), 2)
            draw.text(point_1, values, font=font1, fill=(0, 0, 255))

    img_draw_show= np.array(img_draw)
    cv2.imshow("image",image_cv)
    cv2.imshow("img_draw", img_draw_show)
    cv2.waitKey(0)

if __name__ == "__main__":
    path = "./1.png"
    paddle_ocr(path)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38

结果