!6809 modified psenet for safe

Merge pull request !6809 from 吴书全/psenet0923

model_zoo/official/cv/psenet/README.md

@@ -57,11 +57,15 @@ After installing MindSpore via the official website, you can start training and
 # run distributed training example
 sh scripts/run_distribute_train.sh pretrained_model.ckpt
 
-#setup opencv library
-download pyblind11, opencv3.4,setup opencv3.4
+#download opencv library
+download pyblind11, opencv3.4
 
-#make so file
-run src/ETSNET/pse/Makefile; make libadaptor.so
+#install pyblind11 opencv3.4
+setup pyblind11(install the library by the pip command)
+setup opencv3.4(compile source code install the library)
+
+#enter the path ,run Makefile to product file
+cd ./src/ETSNET/pse/;make
 
 #run test.py
 python test.py --ckpt=pretrained_model.ckpt
@@ -81,10 +85,10 @@ sh scripts/run_eval_ascend.sh
 	├── README.md                           // descriptions about PSENet
 	├── scripts  
 		├── run_distribute_train.sh  		// shell script for distributed
-		└── eval_ic15.sh  					// shell script for evaluation 
+		└── run_eval_ascend.sh  			// shell script for evaluation 
 	├── src  
 		├── __init__.py  
-        ├── generate_hccn_file.py          // creating rank.json
+        ├── generate_hccn_file.py           // creating rank.json
 		├── ETSNET  
 			├── __init__.py  
 			├── base.py                     // convolution and BN operator
@@ -127,7 +131,7 @@ sh scripts/run_distribute_train.sh pretrained_model.ckpt
 ```
 
 The above shell script will run distribute training in the background. You can view the results through the file 
-`device[X]/log`. The loss value will be achieved as follows:
+`device[X]/test_*.log`. The loss value will be achieved as follows:
 
 ```
 # grep "epoch: " device_*/loss.log
@@ -140,6 +144,8 @@ device_1/log:epcoh: 2, step: 40, loss is 0.76629
 ```
 
 ## [Evaluation Process](#contents)
+### run test code
+python test.py --ckpt=./device*/ckpt*/ETSNet-*.ckpt
 
 ### Eval Script for ICDAR2015
 #### Usage
@@ -187,7 +193,7 @@ Calculated!{"precision": 0.814796668299853, "recall": 0.8006740491092923, "hmean
 | MindSpore Version   | 1.0-alpha                   |
 | Dataset             | ICDAR2015                   |
 | outputs             | probability                 |
-| Accuracy            | 1pc: 81%;  8pcs: 81%   |
+| Accuracy            | 1pc: 81%;  4pcs: 81%   |
 
 ## [How to use](#contents)
 

model_zoo/official/cv/psenet/src/ETSNET/pse/adaptor.cpp

@@ -49,7 +49,7 @@ namespace pse_adaptor {
         Mat label_mat;
         int label_num = connectedComponents(kernals[kernals.size() - 1], label_mat, 4);
         vector<int> area(label_num + 1, 0)
-        memset(area, 0, sizeof(area));
+        memset_s(area, 0, sizeof(area));
         for (int x = 0; x < label_mat.rows; ++x) {
             for (int y = 0; y < label_mat.cols; ++y) {
                 int label = label_mat.at<int>(x, y);

model_zoo/official/cv/psenet/src/ETSNET/pse/adaptor.h

@@ -13,3 +13,7 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */
+
+#ifndef ADAPTOR_H
+#define ADAPTOR_H
+#endif

model_zoo/official/cv/psenet/train.py

@@ -49,7 +49,7 @@ def train():
     rank_id = 0
     if args.run_distribute:
         context.set_auto_parallel_context(device_num=args.device_num, parallel_mode=ParallelMode.DATA_PARALLEL,
-                                          gradients_mean=True, parameter_broadcast=True)
+                                          gradients_mean=True)
         init()
         rank_id = get_rank()