python 怎样指定det,rec模型和dict, engine = RapidOCR()

[nissansz]

-- encoding: utf-8 --

@author: SWHL

@Contact: liekkaskono@163.com

import cv2

from rapidocr_onnxruntime import RapidOCR, VisRes

from rapidocr_openvino import RapidOCR, VisRes

engine = RapidOCR()
vis = VisRes(font_path="resources/fonts/FZYTK.TTF")

image_path = "tests/test_files/ch_en_num.jpg"
with open(image_path, "rb") as f:
img = f.read()

result, elapse_list = engine(img)
print(result)
print(elapse_list)

boxes, txts, scores = list(zip(*result))
vis_img = vis(img, boxes, txts, scores)
cv2.imwrite("vis.png", vis_img)

[SWHL]

建议先去看一下官方文档哈

[nissansz]

实验了一张手写的检测，跟ctpn检测没啥区别

…

--------------------------------------------------------------------------------

------------------ 原始邮件 ------------------ 发件人: SWHL ***@***.***> 发送时间: 2023-10-06 17:40:18 收件人:RapidAI/RapidOCR ***@***.***> 抄送:nissanjp ***@***.***>,Author ***@***.***> 主题: Re: [RapidAI/RapidOCR] python 怎样指定det,rec模型和dict, engine = RapidOCR() (Discussion #129) 建议先去看一下官方文档哈发自我的iPhone------------------ 原始邮件 ------------------发件人: nissansz ***@***.***>发送时间: 2023年10月6日 06:39收件人: RapidAI/RapidOCR ***@***.***>抄送: Subscribed ***@***.***>主题: Re: [RapidAI/RapidOCR] python 怎样指定det,rec模型和dict, engine = RapidOCR() (Discussion #129) -- encoding: utf-8 -- @author: SWHL @Contact: ***@***.*** import cv2 from rapidocr_onnxruntime import RapidOCR, VisRes from rapidocr_openvino import RapidOCR, VisRes engine = RapidOCR() vis = VisRes(font_path="resources/fonts/FZYTK.TTF") image_path = "tests/test_files/ch_en_num.jpg" with open(image_path, "rb") as f: img = f.read() result, elapse_list = engine(img) print(result) print(elapse_list) boxes, txts, scores = list(zip(*result)) vis_img = vis(img, boxes, txts, scores) cv2.imwrite("vis.png", vis_img) —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***> [ { ***@***.***": "http://schema.org", ***@***.***": "EmailMessage", "potentialAction": { ***@***.***": "ViewAction", "target": "#129", "url": "#129", "name": "View Discussion" }, "description": "View this Discussion on GitHub", "publisher": { ***@***.***": "Organization", "name": "GitHub", "url": "https://github.com" } } ] — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[SWHL]

我不太清楚你的问题是什么？如果可以详细具体地说明一下你的问题，我这里可以进一步给出参考。

[nissansz]

有没有pixellink, seglink等onnx检测模型？

[SWHL]

没有，目前仓库模型仅限于PaddleOCR中mobile和server两个版本的，其他算法模型因通用性较差，并无涉及。