Skip to content

周伟奇 / part_of_F3_OCR

Go to a project
Toggle navigation
Toggle navigation pinning
98ec226f by 周伟奇
Options

fix auth from

1 parent 7e35e5ba
Inline Side-by-side
Showing 3 changed files with 57 additions and 32 deletions
README.md
...@@ -18,7 +18,7 @@ print(res) # {'label': '营业执照', 'confidence': 0.988462} ...@@ -18,7 +18,7 @@ print(res) # {'label': '营业执照', 'confidence': 0.988462}
18 from authorization_from import retriever_individuals, retriever_companies 18 from authorization_from import retriever_individuals, retriever_companies
19 19
20 # 个人授权书 20 # 个人授权书
21 res = retriever_companies.get_target_fields(go_res, signature_res) 21 res = retriever_individuals.get_target_fields(go_res, signature_res)
22 print(res) 22 print(res)
23 # {'words_result': 23 # {'words_result':
24 # {'姓名': {'words': 'xx', 'score': 1, 'location': {'left': 105, 'top': 277, 'width': 60, 'height': 28}}, 24 # {'姓名': {'words': 'xx', 'score': 1, 'location': {'left': 105, 'top': 277, 'width': 60, 'height': 28}},
...@@ -27,7 +27,7 @@ print(res) ...@@ -27,7 +27,7 @@ print(res)
27 # } 27 # }
28 28
29 # 公司授权书 29 # 公司授权书
30 # res = retriever_individuals.get_target_fields(go_res, signature_res) 30 # res = retriever_companies.get_target_fields(go_res, signature_res)
31 # print(res) 31 # print(res)
32 32
33 # {'words_result': { 33 # {'words_result': {
......
authorization_from/const.py
...@@ -4,8 +4,8 @@ TARGET_FIELD_INDIVIDUALS = { ...@@ -4,8 +4,8 @@ TARGET_FIELD_INDIVIDUALS = {
4 '个人身份证件号码': [('个人身份证件号码', 'top1', {})], 4 '个人身份证件号码': [('个人身份证件号码', 'top1', {})],
5 }, 5 },
6 'value': { 6 'value': {
7 '姓名': ('under', {'left_padding': 1, 'right_padding': 1}, ''), 7 '姓名': ('under', {'left_padding': 1, 'right_padding': 1, 'scope': 2}, ''),
8 '个人身份证件号码': ('under', {'left_padding': 0.5, 'right_padding': 0.5}, '') 8 '个人身份证件号码': ('under', {'left_padding': 0.5, 'right_padding': 0.5, 'scope': 2}, '')
9 }, 9 },
10 'signature': { 10 'signature': {
11 '签字': {'signature', } 11 '签字': {'signature', }
...@@ -19,17 +19,17 @@ TARGET_FIELD_COMPANIES = { ...@@ -19,17 +19,17 @@ TARGET_FIELD_COMPANIES = {
19 ], 19 ],
20 '经销商代码-宝马中国': [ 20 '经销商代码-宝马中国': [
21 ('经销商代码', 'top1', {}), 21 ('经销商代码', 'top1', {}),
22 ('宝马中国', 'right', {'top_padding': 1.5, 'bottom_padding': 0}) 22 ('宝马中国', 'right', {'top_padding': 1.5, 'bottom_padding': 0, 'scope': 2})
23 ], 23 ],
24 '管理人员姓名-总经理': [ 24 '管理人员姓名-总经理': [
25 ('管理人员姓名', 'top1', {}), 25 ('管理人员姓名', 'top1', {}),
26 ('总经理', 'right', {'top_padding': 1, 'bottom_padding': 0}) 26 ('总经理', 'right', {'top_padding': 1, 'bottom_padding': 0, 'scope': 2})
27 ], 27 ],
28 }, 28 },
29 'value': { 29 'value': {
30 '经销商名称': ('right', {'top_padding': 1, 'bottom_padding': 1}, ''), 30 '经销商名称': ('right', {'top_padding': 1, 'bottom_padding': 1, 'scope': 4}, ''),
31 '经销商代码-宝马中国': ('right', {'top_padding': 0.5, 'bottom_padding': 0.5}, ''), 31 '经销商代码-宝马中国': ('right', {'top_padding': 0.5, 'bottom_padding': 0.5, 'scope': 3, 'value_type': 'int'}, ''),
32 '管理人员姓名-总经理': ('right', {'top_padding': 0.5, 'bottom_padding': 0.5}, '') 32 '管理人员姓名-总经理': ('right', {'top_padding': 0.5, 'bottom_padding': 0.5, 'scope': 5}, '')
33 }, 33 },
34 'signature': { 34 'signature': {
35 '公司公章': {'circle', }, 35 '公司公章': {'circle', },
......
authorization_from/retriever.py
...@@ -8,6 +8,11 @@ class Retriever: ...@@ -8,6 +8,11 @@ class Retriever:
8 self.signature_have_not_str = '无' 8 self.signature_have_not_str = '无'
9 self.target_fields = target_fields 9 self.target_fields = target_fields
10 self.key_text_set = self.get_key_text_set(target_fields) 10 self.key_text_set = self.get_key_text_set(target_fields)
11 self.replace_map = {
12 'int': {
13 '(': '0'
14 }
15 }
11 16
12 def get_key_text_set(self, target_fields): 17 def get_key_text_set(self, target_fields):
13 # 关键词集合 18 # 关键词集合
...@@ -24,60 +29,80 @@ class Retriever: ...@@ -24,60 +29,80 @@ class Retriever:
24 return coordinates_list[0] 29 return coordinates_list[0]
25 30
26 @staticmethod 31 @staticmethod
27 def key_right(coordinates_list, key_coordinates, top_padding, bottom_padding): 32 def key_right(coordinates_list, key_coordinates, top_padding, bottom_padding, scope):
28 # 关键词查找方向：右侧 33 # 关键词查找方向：右侧
29 if len(coordinates_list) == 1: 34 if len(coordinates_list) == 1:
30 return coordinates_list[0] 35 return coordinates_list[0]
31 height = key_coordinates[-1] - key_coordinates[1] 36 height = key_coordinates[-1] - key_coordinates[1]
32 y_min = key_coordinates[1] - (top_padding * height) 37 y_min = key_coordinates[1] - (top_padding * height)
33 y_max = key_coordinates[-1] + (bottom_padding * height) 38 y_max = key_coordinates[-1] + (bottom_padding * height)
34 x = key_coordinates[2]
35 39
36 x_min = None 40 width = key_coordinates[2] - key_coordinates[0]
41 x_min = key_coordinates[2]
42 x_max = key_coordinates[2] + (width * scope)
43
44 x_min_find = None
37 key_coordinates = None 45 key_coordinates = None
38 for x0, y0, x1, y1 in coordinates_list: 46 for x0, y0, x1, y1 in coordinates_list:
39 if y0 > y_min and y1 < y_max and x0 > x: 47 cent_x = x0 + ((x1 - x0) / 2)
40 if x_min is None or x0 < x_min: 48 cent_y = y0 + ((y1 - y0) / 2)
41 x_min = x0 49 if x_min < cent_x < x_max and y_min < cent_y < y_max:
50 if x_min_find is None or x0 < x_min_find:
51 x_min_find = x0
42 key_coordinates = (x0, y0, x1, y1) 52 key_coordinates = (x0, y0, x1, y1)
43 return key_coordinates 53 return key_coordinates
44 54
45 @staticmethod 55 def value_right(self, go_res, key_coordinates, top_padding, bottom_padding, scope, value_type=None):
46 def value_right(go_res, key_coordinates, top_padding, bottom_padding):
47 # 字段值查找方向：右侧 56 # 字段值查找方向：右侧
48 height = key_coordinates[-1] - key_coordinates[1] 57 height = key_coordinates[-1] - key_coordinates[1]
49 y_min = key_coordinates[1] - (top_padding * height) 58 y_min = key_coordinates[1] - (top_padding * height)
50 y_max = key_coordinates[-1] + (bottom_padding * height) 59 y_max = key_coordinates[-1] + (bottom_padding * height)
51 x = key_coordinates[2]
52 60
53 x_min = None 61 width = key_coordinates[2] - key_coordinates[0]
62 x_min = key_coordinates[2]
63 x_max = key_coordinates[2] + (width * scope)
64
65 x_min_find = None
54 value = None 66 value = None
55 coordinates = None 67 coordinates = None
56 for (x0, y0, _, _, x1, y1, _, _), text in go_res.values(): 68 for (x0, y0, _, _, x1, y1, _, _), text in go_res.values():
57 if y0 > y_min and y1 < y_max and x0 > x: 69 cent_x = x0 + ((x1 - x0) / 2)
58 if x_min is None or x0 < x_min: 70 cent_y = y0 + ((y1 - y0) / 2)
59 x_min = x0 71 if x_min < cent_x < x_max and y_min < cent_y < y_max:
60 value = text 72 if x_min_find is None or x0 < x_min_find:
61 coordinates = (x0, y0, x1, y1) 73 if len(text.strip()) > 0:
74 x_min_find = x0
75 value = text
76 coordinates = (x0, y0, x1, y1)
77
78 if isinstance(value_type, str) and value_type in self.replace_map and isinstance(value, str):
79 new_value = value.translate(str.maketrans(self.replace_map.get(value_type, {})))
80 return new_value, coordinates
62 return value, coordinates 81 return value, coordinates
63 82
64 @staticmethod 83 @staticmethod
65 def value_under(go_res, key_coordinates, left_padding, right_padding): 84 def value_under(go_res, key_coordinates, left_padding, right_padding, scope, value_type=None):
66 # 字段值查找方向：下方 85 # 字段值查找方向：下方
67 width = key_coordinates[2] - key_coordinates[0] 86 width = key_coordinates[2] - key_coordinates[0]
68 x_min = key_coordinates[0] - (width * left_padding) 87 x_min = key_coordinates[0] - (width * left_padding)
69 x_max = key_coordinates[2] + (width * right_padding) 88 x_max = key_coordinates[2] + (width * right_padding)
70 y = key_coordinates[-1]
71 89
72 y_min = None 90 height = key_coordinates[-1] - key_coordinates[1]
91 y_min = key_coordinates[-1]
92 y_max = key_coordinates[-1] + (height * scope)
93
94 y_min_find = None
73 value = None 95 value = None
74 coordinates = None 96 coordinates = None
75 for (x0, y0, _, _, x1, y1, _, _), text in go_res.values(): 97 for (x0, y0, _, _, x1, y1, _, _), text in go_res.values():
76 if x0 > x_min and x1 < x_max and y0 > y: 98 cent_x = x0 + ((x1 - x0)/2)
77 if y_min is None or y0 < y_min: 99 cent_y = y0 + ((y1 - y0)/2)
78 y_min = y0 100 if x_min < cent_x < x_max and y_min < cent_y < y_max:
79 value = text 101 if y_min_find is None or y0 < y_min_find:
80 coordinates = (x0, y0, x1, y1) 102 if len(text.strip()) > 0:
103 y_min_find = y0
104 value = text
105 coordinates = (x0, y0, x1, y1)
81 return value, coordinates 106 return value, coordinates
82 107
83 @staticmethod 108 @staticmethod
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!