update python/README

This commit is contained in:
SWHL committed 2021-06-09 22:27:45 +08:00
1 parent 69282ed944
commit 6159cd7205
2 files changed
+15 -62

No files matched your search

+14 -61
View File
@@ -1,3 +1,6 @@
### 2021-06-09 update:
- 添加字体文件到模型下载链接中
### 2021-05-22 update:
- 添加日文识别推理代码和相应模型,详情见`japan_ppocr_mobile_v2_rec`目录
- 整理代码,去除无效代码
@@ -6,15 +9,20 @@
### 2021-03-24 update:
- 更改文本识别模型支持ort1.7,推理速度有着较大提升
### 检测、方向分类和识别模型下载:
- [链接](https://pan.baidu.com/s/1qkqWK4wRdMjqGGbzR-FyWg)
- 提取码:30jv
- 下载之后放在`models`目录即可
---
### 运行
1. 下载相应模型和用于显示的字体文件
- [提取码:30jv](https://pan.baidu.com/s/1qkqWK4wRdMjqGGbzR-FyWg)
- 下载之后放在`models`目录即可
2. 运行`rapidOCR.py`文件, 即可
```shell
cd python
python rapidOCR.py
```
### 文本检测+方向分类+文本识别
- 具体参见[rapidOCR.py](./rapidOCR.py)
#### 简单示例
```python
from ch_ppocr_mobile_v2_cls import TextClassifier
from ch_ppocr_mobile_v2_det import TextDetector
@@ -33,58 +41,3 @@ text_sys = TextSystem(det_model_path,
dt_boxes, rec_res = text_sys(image_path)
visualize(image_path, dt_boxes, rec_res)
```
### 通用超轻量文本检测模型 ch_ppocr_mobile_v2_det_train
- **Note: 支持任意尺寸输入**
- **server版文本检测只是采用了ResNet18作为backbone,其他与超轻量无区别**
- 详细见[ch_ppocr_mobile_v2_det/text_detect.py](./ch_ppocr_mobile_v2_det/text_detect.py)
#### 简单示例
```python
det_model_path = 'models/ch_ppocr_mobile_v2_det_train.onnx'
image_path = r'test_images/det_images/1.jpg'
test_detector = TextDetector(det_model_path)
img, flag = check_and_read_gif(image_path)
if not flag:
img = cv2.imread(image_path)
if img is None:
raise ValueError(f"error in loading image:{image_path}")
# dst_boxes: 检测到图像中的文本框坐标,ndarray格式
# (10, 4, 2)→[10个,4个坐标,每个坐标两个点]
dt_boxes, elapse = test_detector(img)
print(dt_boxes.shape)
print(elapse)
```
### 通用超轻量方向分类器模型 ch_ppocr_mobile_v2_cls
- **Note**: 支持`0 180`两个方向分类
- 详细见[ch_ppocr_mobile_v2_cls/text_cls.py](./ch_ppocr_mobile_v2_cls/text_cls.py)
#### 简单示例
```python
text_classifier = TextClassifier(args.cls_model_dir)
img_list, cls_res, predict_time = text_classifier(args.image_dir)
for ino in range(len(img_list)):
print(f"分类结果:{cls_res[ino]}")
```
### 通用超轻量文本识别模型 ch_ppocr_mobile_v2_rec
- **Note: 支持任意尺寸输入**
- 详细见[ch_ppocr_mobile_v2_rec/text_recognize.py](./ch_ppocr_mobile_v2_rec/text_recognize.py)
#### 简单示例:
```python
rec_model_path = r'models/ch_ppocr_mobile_v2.0_rec_pre_infer.onnx'
image_path = r'test_images/rec_images/2021-01-19_13-44-34.png'
text_recongnizer = TextRecognizer(rec_model_path)
rec_res, elapse = text_recongnizer(image_path)
print(f'识别结果:{rec_res}/tcost: {elapse}s')
```
+1 -1
View File
@@ -33,7 +33,7 @@ def check_and_read_gif(img_path):
def draw_ocr_box_txt(image, boxes, txts,
scores=None, drop_score=0.5,
font_path="./doc/simfang.ttf"):
font_path="./models/msyh.ttc"):
h, w = image.height, image.width
img_left = image.copy()
img_right = Image.new('RGB', (w, h), (255, 255, 255))