fix bug

doc/doc_ch/algorithm_overview.md

@@ -107,4 +107,4 @@
 
 |模型|骨干网络|配置文件|acc|下载链接|
 |---|---|---|---|---|
-|TableMaster|TableResNetExtra|[configs/table/table_master.yml](../../configs/table/table_master.yml)|77.47%|[训练模型]|[训练模型](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_train.tar)/[推理模型](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_infer.tar)|
+|TableMaster|TableResNetExtra|[configs/table/table_master.yml](../../configs/table/table_master.yml)|77.47%|[训练模型](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_train.tar) / [推理模型](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_infer.tar)|

doc/doc_ch/algorithm_table_master.md

@@ -55,7 +55,7 @@ python3 tools/export_model.py -c configs/table/table_master.yml -o Global.pretra
 
 转换成功后，在目录下有三个文件：
 ```
-/inference/table_master/
+./inference/table_master/
     ├── inference.pdiparams         # 识别inference模型的参数文件
     ├── inference.pdiparams.info    # 识别inference模型的参数信息，可忽略
     └── inference.pdmodel           # 识别inference模型的program文件

doc/doc_en/algorithm_overview_en.md

@@ -1,10 +1,10 @@
 # OCR Algorithms
 
-- [1. Two-stage Algorithms](#1)
-  * [1.1 Text Detection Algorithms](#11)
-  * [1.2 Text Recognition Algorithms](#12)
-- [2. End-to-end Algorithms](#2)
-- [3. Table Recognition Algorithms](#3)
+- [1. Two-stage Algorithms](#1-two-stage-algorithms)
+  - [1.1 Text Detection Algorithms](#11-text-detection-algorithms)
+  - [1.2 Text Recognition Algorithms](#12-text-recognition-algorithms)
+- [2. End-to-end Algorithms](#2-end-to-end-algorithms)
+- [3. Table Recognition Algorithms](#3-table-recognition-algorithms)
 
 
 This tutorial lists the OCR algorithms supported by PaddleOCR, as well as the models and metrics of each algorithm on **English public datasets**. It is mainly used for algorithm introduction and algorithm performance comparison. For more models on other datasets including Chinese, please refer to [PP-OCR v2.0 models list](./models_list_en.md).
@@ -107,4 +107,4 @@ On the PubTabNet dataset, the algorithm result is as follows:
 
 |Model|Backbone|Config|Acc|Download link|
 |---|---|---|---|---|
-|TableMaster|TableResNetExtra|[configs/table/table_master.yml](../../configs/table/table_master.yml)|77.47%|[训练模型]|[训练模型](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_train.tar)/[推理模型](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_infer.tar)|
+|TableMaster|TableResNetExtra|[configs/table/table_master.yml](../../configs/table/table_master.yml)|77.47%|[trained](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_train.tar) / [inference model](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_infer.tar)|

doc/doc_en/algorithm_table_master_en.md

@@ -1,4 +1,4 @@
-# Torm Recognition Algorithm-TableMASTER
+# Table Recognition Algorithm-TableMASTER
 
 - [1. Introduction](#1-introduction)
 - [2. Environment](#2-environment)
@@ -24,7 +24,7 @@ On the PubTabNet table recognition public data set, the algorithm reproduction a
 
 |Model|Backbone|Cnnfig|Acc|Download link|
 | --- | --- | --- | --- | --- |
-|TableMaster|TableResNetExtra|[configs/table/table_master.yml](../../configs/table/table_master.yml)|77.47%|[train model](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_train.tar)/[inference model](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_infer.tar)|
+|TableMaster|TableResNetExtra|[configs/table/table_master.yml](../../configs/table/table_master.yml)|77.47%|[trained model](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_train.tar)/[inference model](https://paddleocr.bj.bcebos.com/ppstructure/models/tablemaster/table_structure_tablemaster_infer.tar)|
 
 
 <a name="2"></a>

ppocr/data/imaug/label_ops.py

@@ -671,7 +671,7 @@ class TableLabelEncode(AttnLabelEncode):
 
     def _merge_no_span_structure(self, structure):
         """
-        This fun code is refer from:
+        This code is refer from:
         https://github.com/JiaquanYe/TableMASTER-mmocr/blob/master/table_recognition/data_preprocess.py
         """
         new_structure = []

ppocr/losses/table_master_loss.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """
-This fun code is refer from:
+This code is refer from:
 https://github.com/JiaquanYe/TableMASTER-mmocr/tree/master/mmocr/models/textrecog/losses
 """
 

ppocr/metrics/table_metric.py

@@ -31,8 +31,6 @@ class TableStructureMetric(object):
                                              gt_structure_batch_list):
             pred_str = ''.join(pred)
             target_str = ''.join(target)
-            # pred_str = pred_str.replace('<thead>','').replace('</thead>','').replace('<tbody>','').replace('</tbody>','')
-            # target_str = target_str.replace('<thead>','').replace('</thead>','').replace('<tbody>','').replace('</tbody>','')
             if pred_str == target_str:
                 correct_num += 1
             all_num += 1
@@ -55,8 +53,6 @@ class TableStructureMetric(object):
         self.len_acc_num = 0
         self.token_nums = 0
         self.anys_dict = dict()
-        from collections import defaultdict
-        self.error_num_dict = defaultdict(int)
 
 
 class TableMetric(object):

ppocr/modeling/backbones/table_master_resnet.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """
-This fun code is refer from:
+This code is refer from:
 https://github.com/JiaquanYe/TableMASTER-mmocr/blob/master/mmocr/models/textrecog/backbones/table_resnet_extra.py
 """
 
@@ -193,47 +193,43 @@ class TableResNetExtra(nn.Layer):
 
     def forward(self, x):
         f = []
-        x = self.conv1(x)  # 1,64,480,480
+        x = self.conv1(x)
 
         x = self.bn1(x)
         x = self.relu1(x)
 
-        x = self.conv2(x)  # 1,128,480,480
+        x = self.conv2(x)
         x = self.bn2(x)
         x = self.relu2(x)
-        # (48, 160)
 
-        x = self.maxpool1(x)  # 1,64,240,240
+        x = self.maxpool1(x)
         x = self.layer1(x)
 
-        x = self.conv3(x)  # 1,256,240,240
+        x = self.conv3(x)
         x = self.bn3(x)
         x = self.relu3(x)
         f.append(x)
-        # (24, 80)
 
-        x = self.maxpool2(x)  # 1,256,120,120
+        x = self.maxpool2(x)
         x = self.layer2(x)
 
-        x = self.conv4(x)  # 1,256,120,120
+        x = self.conv4(x)
         x = self.bn4(x)
         x = self.relu4(x)
         f.append(x)
-        # (12, 40)
 
-        x = self.maxpool3(x)  # 1,256,60,60
+        x = self.maxpool3(x)
 
-        x = self.layer3(x)  # 1,512,60,60
-        x = self.conv5(x)  # 1,512,60,60
+        x = self.layer3(x)
+        x = self.conv5(x)
         x = self.bn5(x)
         x = self.relu5(x)
 
-        x = self.layer4(x)  # 1,512,60,60
-        x = self.conv6(x)  # 1,512,60,60
+        x = self.layer4(x)
+        x = self.conv6(x)
         x = self.bn6(x)
         x = self.relu6(x)
         f.append(x)
-        # (6, 40)
         return f
 
 

ppocr/modeling/heads/table_master_head.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """
-This fun code is refer from:
+This code is refer from:
 https://github.com/JiaquanYe/TableMASTER-mmocr/blob/master/mmocr/models/textrecog/decoders/master_decoder.py
 """
 
@@ -135,7 +135,7 @@ class TableMasterHead(nn.Layer):
         batch_size = out_enc.shape[0]
         SOS = paddle.zeros([batch_size, 1], dtype='int64') + self.SOS
         output, bbox_output = self.greedy_forward(SOS, out_enc)
-        # output = F.softmax(output)
+        output = F.softmax(output)
         return {'structure_probs': output, 'loc_preds': bbox_output}
 
     def forward(self, feat, targets=None):

ppocr/utils/visual.py

@@ -110,3 +110,16 @@ def draw_re_results(image,
 
     img_new = Image.blend(image, img_new, 0.5)
     return np.array(img_new)
+
+
+def draw_rectangle(img_path, boxes, use_xywh=False):
+    img = cv2.imread(img_path)
+    img_show = img.copy()
+    for box in boxes.astype(int):
+        if use_xywh:
+            x, y, w, h = box
+            x1, y1, x2, y2 = x - w // 2, y - h // 2, x + w // 2, y + h // 2
+        else:
+            x1, y1, x2, y2 = box
+        cv2.rectangle(img_show, (x1, y1), (x2, y2), (255, 0, 0), 2)
+    return img_show

ppstructure/table/predict_structure.py

@@ -30,6 +30,7 @@ from ppocr.data import create_operators, transform
 from ppocr.postprocess import build_post_process
 from ppocr.utils.logging import get_logger
 from ppocr.utils.utility import get_image_file_list, check_and_read_gif
+from ppocr.utils.visual import draw_rectangle
 from ppstructure.utility import parse_args
 
 logger = get_logger()
@@ -120,19 +121,6 @@ class TableStructurer(object):
         return structure_str_list, bbox_list, elapse
 
 
-def draw_rectangle(img_path, boxes, use_xywh=False):
-    img = cv2.imread(img_path)
-    img_show = img.copy()
-    for box in boxes.astype(int):
-        if use_xywh:
-            x, y, w, h = box
-            x1, y1, x2, y2 = x - w // 2, y - h // 2, x + w // 2, y + h // 2
-        else:
-            x1, y1, x2, y2 = box
-        cv2.rectangle(img_show, (x1, y1), (x2, y2), (255, 0, 0), 2)
-    return img_show
-
-
 def main(args):
     image_file_list = get_image_file_list(args.image_dir)
     table_structurer = TableStructurer(args)

tools/infer_table.py

@@ -36,6 +36,7 @@ from ppocr.modeling.architectures import build_model
 from ppocr.postprocess import build_post_process
 from ppocr.utils.save_load import load_model
 from ppocr.utils.utility import get_image_file_list
+from ppocr.utils.visual import draw_rectangle
 import tools.program as program
 import cv2
 
@@ -111,19 +112,6 @@ def main(config, device, logger, vdl_writer):
         logger.info("success!")
 
 
-def draw_rectangle(img_path, boxes, use_xywh=False):
-    img = cv2.imread(img_path)
-    img_show = img.copy()
-    for box in boxes.astype(int):
-        if use_xywh:
-            x, y, w, h = box
-            x1, y1, x2, y2 = x - w // 2, y - h // 2, x + w // 2, y + h // 2
-        else:
-            x1, y1, x2, y2 = box
-        cv2.rectangle(img_show, (x1, y1), (x2, y2), (255, 0, 0), 2)
-    return img_show
-
-
 if __name__ == '__main__':
     config, device, logger, vdl_writer = program.preprocess()
     main(config, device, logger, vdl_writer)