Skip to content

Commit d4c266a

Browse files
committed
chore: update readme
1 parent 7e6a210 commit d4c266a

File tree

2 files changed

+23
-18
lines changed

2 files changed

+23
-18
lines changed

README.md

Lines changed: 21 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -42,6 +42,26 @@
4242
# 建议使用清华源安装 https://pypi.tuna.tsinghua.edu.cn/simple
4343
pip install rapid-table-det
4444
```
45+
46+
#### 参数说明
47+
cpu和gpu的初始化完全一致
48+
```
49+
table_det = TableDetector(
50+
# 目标检测表格模型
51+
obj_model_path="models/obj_det_paddle(obj_det.onnx)",
52+
# 边角检测表格模型(从复杂环境得到表格多边形框)
53+
edge_model_path="models/edge_det_paddle(edge_det.onnx)",
54+
# 角点方向识别
55+
cls_model_path="models/cls_det_paddle(cls_det.onnx)",
56+
# 文档场景已经由版面识别模型提取设置为False
57+
use_obj_det=True,
58+
# 只有90,180,270大角度旋转且无透视时候设置为False
59+
use_edge_det=True,
60+
# 小角度(-90~90)旋转设置为False
61+
use_cls_det=True,
62+
)
63+
```
64+
4565
### 快速使用
4666
``` python {linenos=table}
4767
from rapid_table_det.inference import TableDetector
@@ -117,23 +137,7 @@ print(
117137
# cv2.imwrite(f"{out_dir}/{file_name}-visualize.jpg", img)
118138

119139
```
120-
#### 参数说明
121-
```
122-
table_det = TableDetector(\
123-
# 目标检测表格模型
124-
obj_model_path="models/obj_det_paddle(obj_det.onnx)", \
125-
# 边角检测表格模型(从复杂环境得到表格多边形框)
126-
edge_model_path="models/edge_det_paddle(edge_det.onnx)", \
127-
# 角点方向识别
128-
cls_model_path="models/cls_det_paddle(cls_det.onnx)", \
129-
# 文档场景已经由版面识别模型提取可以不使用
130-
use_obj_det=True, \
131-
# 只有90,180,270大角度旋转且无透视时候可以不使用
132-
use_edge_det=True, \
133-
# 小角度(-90~90)旋转可以不使用
134-
use_cls_det=True, \
135-
)
136-
```
140+
137141

138142
## FAQ (Frequently Asked Questions)
139143

demo_onnx.py

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,9 +1,10 @@
11
from rapid_table_det.inference import TableDetector
22

3-
img_path = f"tests/test_files/chip2.jpg"
3+
img_path = f"tests/test_files/WechatIMG943.jpg"
44
table_det = TableDetector(
55
obj_model_path="rapid_table_det/models/obj_det.onnx",
66
edge_model_path="rapid_table_det/models/edge_det.onnx",
7+
use_edge_det=False, use_cls_det=False
78
)
89
result, elapse = table_det(img_path)
910
obj_det_elapse, edge_elapse, rotate_det_elapse = elapse

0 commit comments

Comments
 (0)