diff --git a/tensorflow_toolkit/lpr/README.md b/tensorflow_toolkit/lpr/README.md
index 6d41634d9f..3476d0db7c 100644
--- a/tensorflow_toolkit/lpr/README.md
+++ b/tensorflow_toolkit/lpr/README.md
@@ -129,9 +129,10 @@ To run the model via OpenVINO one has to freeze TensorFlow graph and
 then convert it to OpenVINO Internal Representation (IR) using Model Optimizer:
 
 ```Bash
-python3 tools/export.py --data_type FP32 chinese_lp/config.py
+python3 tools/export.py --data_type FP32 --output_dir <export_path> chinese_lp/config.py
 ```
 
+**default export path**:
 `lpr/model/export_<step>/frozen_graph` - path to frozen graph
 `lpr/model/export_<step>/IR/<data_type>` - path to converted model in IR format
 
diff --git a/tensorflow_toolkit/lpr/tools/export.py b/tensorflow_toolkit/lpr/tools/export.py
index bc7edf1304..daf931f6b1 100644
--- a/tensorflow_toolkit/lpr/tools/export.py
+++ b/tensorflow_toolkit/lpr/tools/export.py
@@ -29,6 +29,7 @@
 def parse_args():
   parser = argparse.ArgumentParser(description='Export model in IE format')
   parser.add_argument('--data_type', default='FP32', choices=['FP32', 'FP16'], help='Data type of IR')
+  parser.add_argument('--output_dir', default=None, help='Output Directory')
   parser.add_argument('--checkpoint', default=None, help='Default: latest')
   parser.add_argument('path_to_config', help='Path to a config.py')
   return parser.parse_args()
@@ -71,7 +72,7 @@ def main(_):
     raise FileNotFoundError(str(checkpoint))
 
   step = checkpoint.split('.')[-2].split('-')[-1]
-  output_dir = os.path.join(config.model_dir, 'export_{}'.format(step))
+  output_dir = args.output_dir if args.output_dir else os.path.join(config.model_dir, 'export_{}'.format(step))
 
   # Freezing graph
   frozen_dir = os.path.join(output_dir, 'frozen_graph')
diff --git a/tensorflow_toolkit/ssd_detector/README.md b/tensorflow_toolkit/ssd_detector/README.md
index cc4ab0eede..f8cb604334 100644
--- a/tensorflow_toolkit/ssd_detector/README.md
+++ b/tensorflow_toolkit/ssd_detector/README.md
@@ -148,10 +148,11 @@ To run the model via OpenVINO one has to freeze TensorFlow graph and
 then convert it to OpenVINO Internal Representation (IR) using Model Optimizer:
 
 ```
-python3 tools/export.py --data_type FP32 vlp/config.py
+python3 tools/export.py --data_type FP32 --output_dir <export_path> vlp/config.py
 ```
 
-As a result, you'll find three new artifacts:
+As a result, you'll find three new artifacts:  
+**default export path**
 - `vlp/model/export_<step>/frozen_graph/` - path to frozen graph
 - `vlp/model/export_<step>/IR/<data_type>/` - path to converted model in IR format
 
diff --git a/tensorflow_toolkit/ssd_detector/tools/export.py b/tensorflow_toolkit/ssd_detector/tools/export.py
index d8cf5d571e..a8418d2d6d 100644
--- a/tensorflow_toolkit/ssd_detector/tools/export.py
+++ b/tensorflow_toolkit/ssd_detector/tools/export.py
@@ -28,6 +28,7 @@ def parse_args():
   parser = argparse.ArgumentParser(description='Export model in IE format')
   parser.add_argument('--model_name', default='vlp')
   parser.add_argument('--data_type', default='FP32', choices=['FP32', 'FP16'], help='Data type of IR')
+  parser.add_argument('--output_dir', default=None, help='Output Directory')
   parser.add_argument('--checkpoint', default=None, help='Default: latest')
   parser.add_argument('path_to_config', help='Path to a config.py')
   return parser.parse_args()
@@ -88,7 +89,7 @@ def main(_):
     raise FileNotFoundError(str(checkpoint))
 
   step = checkpoint.split('-')[-1]
-  output_dir = os.path.join(config.MODEL_DIR, 'export_{}'.format(step))
+  output_dir = args.output_dir if args.output_dir else os.path.join(config.MODEL_DIR, 'export_{}'.format(step))
 
   # Freezing graph
   frozen_dir = os.path.join(output_dir, 'frozen_graph')
diff --git a/tensorflow_toolkit/vehicle_attributes/README.md b/tensorflow_toolkit/vehicle_attributes/README.md
index 24c9cb5c1e..6712ca8e6b 100644
--- a/tensorflow_toolkit/vehicle_attributes/README.md
+++ b/tensorflow_toolkit/vehicle_attributes/README.md
@@ -122,10 +122,10 @@ To run the model via OpenVINO one has to freeze TensorFlow graph and
 then convert it to OpenVINO Internal Representation (IR) using Model Optimizer:
 
 ```Bash
-python3 tools/export.py --data_type FP32 cars_100/config.py
+python3 tools/export.py --data_type FP32 --output_dir <export_path> cars_100/config.py
 ```
 
-As a result, you'll find three new artifacts:
+**default export path**
 - `lpr/model/export_<step>/frozen_graph/` - path to frozen graph
 - `lpr/model/export_<step>/IR/<data_type>/` - path to converted model in IR format
 
diff --git a/tensorflow_toolkit/vehicle_attributes/tools/export.py b/tensorflow_toolkit/vehicle_attributes/tools/export.py
index c4b8d4a10f..9535cb1723 100644
--- a/tensorflow_toolkit/vehicle_attributes/tools/export.py
+++ b/tensorflow_toolkit/vehicle_attributes/tools/export.py
@@ -28,6 +28,7 @@ def parse_args():
   parser.add_argument('--mo', default='mo.py', help="Path to model optimizer 'mo.py' script")
   parser.add_argument('--mo_config', default='cars_100/mo.yaml', help="Path config for model optimizer")
   parser.add_argument('--data_type', default='FP32', choices=['FP32', 'FP16'], help='Data type of IR')
+  parser.add_argument('--output_dir', default=None, help='Output Directory')
   parser.add_argument('--checkpoint', default=None, help='Default: latest')
   parser.add_argument('path_to_config', help='Path to a config.py')
   return parser.parse_args()
@@ -59,7 +60,7 @@ def main(_):
     raise FileNotFoundError(str(checkpoint))
 
   step = checkpoint.split('.')[-1].split('-')[-1]
-  output_dir = os.path.join(config.model_dir, 'export_{}'.format(step))
+  output_dir = args.output_dir if args.output_dir else os.path.join(config.model_dir, 'export_{}'.format(step))
 
   # Freezing graph
   frozen_dir = os.path.join(output_dir, 'frozen_graph')