!6611 Add random seed in deeplabv3

Merge pull request !6611 from jiangzhenguang/add_set_random_in_deeplabv3

model_zoo/official/cv/deeplabv3/README.md

@@ -15,6 +15,7 @@
 - [Model Description](#model-description)
     - [Performance](#performance)
         - [Evaluation Performance](#evaluation-performance)
+- [Description of Random Situation](#description-of-random-situation)
 - [ModelZoo Homepage](#modelzoo-homepage)
 
 
@@ -141,7 +142,7 @@ run_eval_s8_multiscale_flip.sh
     ├── run_standalone_train.sh                   # launch ascend standalone training(1 pc)
   ├── src
     ├── data                          
-        ├── data_generator.py                     # mindrecord data generator
+        ├── dataset.py                            # mindrecord data generator
         ├── build_seg_data.py                     # data preprocessing
     ├── loss
        ├── loss.py                                # loss definition for deeplabv3        
@@ -412,10 +413,9 @@ Note: There OS is output stride, and MS is multiscale.
 | Checkpoint for Fine tuning | 443M (.ckpt file)                                         |
 | Scripts                    | [Link](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/official/cv/deeplabv3) | 
 
+# [Description of Random Situation](#contents)
+In dataset.py, we set the seed inside "create_dataset" function. We also use random seed in train.py. 
+
 # [ModelZoo Homepage](#contents)
  Please check the official [homepage](https://gitee.com/mindspore/mindspore/tree/master/model_zoo).
-
+ 
-
-
-
-

model_zoo/official/cv/deeplabv3/train.py

@@ -24,10 +24,13 @@ from mindspore.train.serialization import load_checkpoint, load_param_into_net
 from mindspore.communication.management import init, get_rank, get_group_size
 from mindspore.train.callback import LossMonitor, TimeMonitor
 from mindspore.train.loss_scale_manager import FixedLossScaleManager
-from src.data import data_generator
+from mindspore.common import set_seed
+from src.data import dataset as data_generator
 from src.loss import loss
 from src.nets import net_factory
 from src.utils import learning_rates
+
+set_seed(1)
 context.set_context(mode=context.GRAPH_MODE, enable_auto_mixed_precision=True, save_graphs=False,
                     device_target="Ascend", device_id=int(os.getenv('DEVICE_ID')))