add Semantic Segmentation to registry

Author: Ragav Venkatesan

CHANGELOG.rst

@@ -8,6 +8,7 @@ CHANGELOG
 * bug-fix: Fix FileNotFoundError for entry_point without source_dir
 * doc-fix: Add missing feature 1.5.0 in change log
 * doc-fix: Add README for airflow
+* feature: Add Amazon SageMaker Semantic Segmentation algorithm to the registry.
 
 1.15.1
 ======
@@ -363,4 +364,4 @@ CHANGELOG
 1.0.0
 =====
 
-* Initial commit
+* Initial commit

README.rst

@@ -344,7 +344,7 @@ Currently, the following algorithms support incremental training:
 
 - Image Classification
 - Object Detection
-- Semantics Segmentation
+- Semantic Segmentation
 
 
 MXNet SageMaker Estimators

src/sagemaker/amazon/amazon_estimator.py

@@ -336,7 +336,7 @@ def registry(region_name, algorithm=None):
             "us-west-1": "632365934929",
         }[region_name]
     elif algorithm in ["xgboost", "seq2seq", "image-classification", "blazingtext",
-                       "object-detection"]:
+                       "object-detection", "semantic-segmentation"]:
         account_id = {
             "us-east-1": "811284229777",
             "us-east-2": "825641698319",

tests/integ/test_inference_pipeline.py

@@ -75,10 +75,12 @@ def test_inference_pipeline_model_deploy(sagemaker_session):
         }
     })
     with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
-        sparkml_model = SparkMLModel(model_data=sparkml_model_data, env={'SAGEMAKER_SPARKML_SCHEMA': schema})
+        sparkml_model = SparkMLModel(model_data=sparkml_model_data, env={'SAGEMAKER_SPARKML_SCHEMA': schema},
+                                     sagemaker_session=sagemaker_session)
         xgb_image = get_image_uri(Session().boto_region_name, 'xgboost')
-        xgb_model = Model(model_data=xgb_model_data, image=xgb_image)
-        model = PipelineModel(models=[sparkml_model, xgb_model], role='SageMakerRole', name=endpoint_name)
+        xgb_model = Model(model_data=xgb_model_data, image=xgb_image, sagemaker_session=sagemaker_session)
+        model = PipelineModel(models=[sparkml_model, xgb_model], role='SageMakerRole',
+                              sagemaker_session=sagemaker_session, name=endpoint_name)
         model.deploy(1, 'ml.m4.xlarge', endpoint_name=endpoint_name)
         predictor = RealTimePredictor(endpoint=endpoint_name, sagemaker_session=sagemaker_session,
                                       serializer=json_serializer, content_type=CONTENT_TYPE_CSV,

tests/integ/test_sparkml_serving.py

@@ -65,7 +65,8 @@ def test_sparkml_model_deploy(sagemaker_session):
         }
     })
     with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
-        model = SparkMLModel(model_data=model_data, role='SageMakerRole', env={'SAGEMAKER_SPARKML_SCHEMA': schema})
+        model = SparkMLModel(model_data=model_data, role='SageMakerRole', sagemaker_session=sagemaker_session,
+                             env={'SAGEMAKER_SPARKML_SCHEMA': schema})
         predictor = model.deploy(1, 'ml.m4.xlarge', endpoint_name=endpoint_name)
 
         valid_data = "1.0,C,38.0,71.5,1.0,female"

tests/integ/test_tf.py

@@ -162,4 +162,4 @@ def test_failed_tf_training(sagemaker_session, tf_full_version):
 
         with pytest.raises(ValueError) as e:
             estimator.fit()
-        assert 'ExecuteUserScriptError' in str(e.value)
+        assert 'This failure is expected' in str(e.value)