Spaces:
Running
Running
breezedeus
commited on
Commit
•
877c27b
1
Parent(s):
c305175
Upload app.py
Browse files
app.py
ADDED
@@ -0,0 +1,121 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding: utf-8
|
2 |
+
# Copyright (C) 2022, [Breezedeus](https://github.com/breezedeus).
|
3 |
+
|
4 |
+
from PIL import Image, ImageFilter
|
5 |
+
import streamlit as st
|
6 |
+
|
7 |
+
from cnocr import CnOcr
|
8 |
+
from cnocr.utils import set_logger
|
9 |
+
|
10 |
+
from antiocr.anti_ocr import AntiOcr
|
11 |
+
from antiocr.consts import RESOURCE_PATH
|
12 |
+
|
13 |
+
logger = set_logger()
|
14 |
+
st.set_page_config(layout="wide")
|
15 |
+
|
16 |
+
FONT_FP = '/System/Library/Fonts/PingFang.ttc'
|
17 |
+
|
18 |
+
|
19 |
+
@st.cache(allow_output_mutation=True)
|
20 |
+
def get_ocr_model():
|
21 |
+
return CnOcr()
|
22 |
+
|
23 |
+
|
24 |
+
def main():
|
25 |
+
st.sidebar.header('输出设置')
|
26 |
+
char_reverse_ratio = st.sidebar.slider(
|
27 |
+
'文字倒转概率', min_value=0.0, max_value=1.0, value=0.1
|
28 |
+
)
|
29 |
+
char_to_pinyin_ratio = st.sidebar.slider(
|
30 |
+
'文字转拼音概率', min_value=0.0, max_value=1.0, value=0.1
|
31 |
+
)
|
32 |
+
cols = st.sidebar.columns(2)
|
33 |
+
min_font_size = int(cols[0].number_input('最小文字大小', 2, 80, value=15))
|
34 |
+
max_font_size = int(
|
35 |
+
cols[1].number_input(
|
36 |
+
'最大文字大小', min_font_size + 1, 120, value=max(40, min_font_size + 1)
|
37 |
+
)
|
38 |
+
)
|
39 |
+
text_color = st.sidebar.color_picker('文字颜色', value='#5087DC')
|
40 |
+
|
41 |
+
st.sidebar.markdown('----')
|
42 |
+
use_random_bg = st.sidebar.checkbox('随机生成背景图片')
|
43 |
+
if use_random_bg:
|
44 |
+
bg_text_density = st.sidebar.slider(
|
45 |
+
'背景图片文字密度', min_value=0.0, max_value=3.0, value=1.0
|
46 |
+
)
|
47 |
+
cols = st.sidebar.columns(2)
|
48 |
+
bg_min_size = int(
|
49 |
+
cols[0].number_input('背景图片最小文字', 2, 80, key='bg_min', value=15)
|
50 |
+
)
|
51 |
+
bg_max_size = int(
|
52 |
+
cols[1].number_input(
|
53 |
+
'背景图片最大文字',
|
54 |
+
bg_min_size + 1,
|
55 |
+
120,
|
56 |
+
key='bg_max',
|
57 |
+
value=max(70, bg_min_size + 1),
|
58 |
+
)
|
59 |
+
)
|
60 |
+
bg_text_color = st.sidebar.color_picker('背景图片文字颜色', value='#07BCE0')
|
61 |
+
bg_gen_config = dict(
|
62 |
+
text_density=bg_text_density,
|
63 |
+
text_color=bg_text_color,
|
64 |
+
min_font_size=bg_min_size,
|
65 |
+
max_font_size=bg_max_size,
|
66 |
+
)
|
67 |
+
bg_image = None
|
68 |
+
else:
|
69 |
+
bg_gen_config = None
|
70 |
+
bg_image = Image.open(RESOURCE_PATH / 'bg.jpeg')
|
71 |
+
bg_image = bg_image.filter(ImageFilter.MaxFilter(3))
|
72 |
+
|
73 |
+
title = '让文字自由传播:<a href="https://github.com/breezedeus/antiOCR">antiOCR</a>'
|
74 |
+
st.markdown(f"<h1 style='text-align: center;'>{title}</h1>", unsafe_allow_html=True)
|
75 |
+
subtitle = '作者:<a href="https://github.com/breezedeus">breezedeus</a>; ' \
|
76 |
+
'欢迎加入 <a href="https://cnocr.readthedocs.io/zh/latest/contact/">交流群</a>'
|
77 |
+
st.markdown(f"<div style='text-align: center;'>{subtitle}</div>", unsafe_allow_html=True)
|
78 |
+
st.markdown('')
|
79 |
+
st.markdown('')
|
80 |
+
desc = '<strong>antiOCR</strong> 对指定的文字(来自输入或者图片)进行处理,输出图片,此图片无法通过OCR技术识别出有意义的文字。'
|
81 |
+
st.markdown(f"<div style='text-align: left;'>{desc}</div>", unsafe_allow_html=True)
|
82 |
+
st.markdown('')
|
83 |
+
st.subheader('选择待转换文字图片,或者直接输入待转换文字')
|
84 |
+
default_texts = '真的猛士,敢于直面惨淡的人生,敢于正视淋漓的鲜血。这是怎样的哀痛者和幸福者?然而造化又常常为庸人设计,以时间的流逝,来洗涤旧迹,仅是留下淡红的血色和微漠的悲哀。在这淡红的血色和微漠的悲哀中,又给人暂得偷生,维持着这似人非人的世界。 ——鲁迅'
|
85 |
+
content_file = st.file_uploader('待转换文字图片', type=["png", "jpg", "jpeg", "webp"])
|
86 |
+
ocr = get_ocr_model()
|
87 |
+
anti = AntiOcr()
|
88 |
+
texts = None
|
89 |
+
if content_file is not None:
|
90 |
+
try:
|
91 |
+
img = Image.open(content_file).convert('RGB')
|
92 |
+
ocr_out = ocr.ocr(img)
|
93 |
+
texts = '\n'.join([out['text'] for out in ocr_out])
|
94 |
+
except Exception as e:
|
95 |
+
st.error(e)
|
96 |
+
|
97 |
+
texts = st.text_area('待转换文字图片', value=texts or default_texts, height=120)
|
98 |
+
|
99 |
+
if st.button("生成图片"):
|
100 |
+
if texts:
|
101 |
+
with st.spinner('图片生成中…'):
|
102 |
+
out_img = anti(
|
103 |
+
texts,
|
104 |
+
char_reverse_ratio=char_reverse_ratio,
|
105 |
+
char_to_pinyin_ratio=char_to_pinyin_ratio,
|
106 |
+
text_color=text_color,
|
107 |
+
min_font_size=min_font_size,
|
108 |
+
max_font_size=max_font_size,
|
109 |
+
bg_image=bg_image,
|
110 |
+
bg_gen_config=bg_gen_config,
|
111 |
+
)
|
112 |
+
st.subheader('输出图片')
|
113 |
+
st.image(out_img)
|
114 |
+
st.markdown('**对输出图片进行OCR,结果如下(如果依旧出现敏感词,尝试重新生成图片):**')
|
115 |
+
ocr_out = ocr.ocr(out_img)
|
116 |
+
new_texts = [out['text'] for out in ocr_out]
|
117 |
+
st.text('\n'.join(new_texts))
|
118 |
+
|
119 |
+
|
120 |
+
if __name__ == '__main__':
|
121 |
+
main()
|