diff --git a/sklearn2pmml/__init__.py b/sklearn2pmml/__init__.py
index 07eb98f..82857a2 100644
--- a/sklearn2pmml/__init__.py
+++ b/sklearn2pmml/__init__.py
@@ -1,4 +1,5 @@
 from pandas import CategoricalDtype
+from pathlib import Path
 try:
 	from sklearn_pandas import DataFrameMapper
 except ImportError:
@@ -242,7 +243,7 @@ def sklearn2pmml(estimator, pmml_path, with_repr = False, pmml_schema = None, ja
 
 	Parameters:
 	----------
-	estimator: BaseEstimator
+	estimator: BaseEstimator or path-like
 		The estimator or pipeline object.
 
 	pmml_path: string
@@ -285,36 +286,47 @@ def sklearn2pmml(estimator, pmml_path, with_repr = False, pmml_schema = None, ja
 		print("dill: {0}".format(dill.__version__))
 		print("joblib: {0}".format(joblib.__version__))
 		print("{0}: {1}".format(java_version[0], java_version[1]))
-	if not _is_supported(estimator):
-		raise TypeError("The estimator object is not an instance of {0}".format(BaseEstimator.__name__))
-	# if isinstance(estimator, Pipeline):
-	if hasattr(estimator, "_final_estimator"):
-		final_estimator = estimator._final_estimator
-	else:
-		final_estimator = estimator
+
 	dumps = []
 	try:
-		java_args = ["-cp", os.pathsep.join(_classpath(user_classpath)), "com.sklearn2pmml.Main"]
-		if with_repr:
-			estimator.repr_ = repr(estimator)
-		# if isinstance(final_estimator, H2OEstimator):
-		if hasattr(final_estimator, "download_mojo"):
-			if dump_flavour != "dill":
-				warnings.warn("Changing dump flavour to dill")
-				dump_flavour = "dill"
-			# Avoid MOJO (re-)download if the indicator attribute is set
-			if not (hasattr(final_estimator, "_mojo_path") or hasattr(final_estimator, "_mojo_bytes")):
-				mojo_path = final_estimator.download_mojo()
-				dumps.append(mojo_path)
-				final_estimator._mojo_path = mojo_path
-		if dump_flavour == "dill":
-			pkl_path = _dill_dump(estimator, "estimator")
-		elif dump_flavour == "joblib":
-			pkl_path = _joblib_dump(estimator, "estimator")
+		if isinstance(estimator, (str, Path)):
+			if with_repr:
+				warnings.warn("Ignoring 'with_repr' flag")
+
+			pkl_path = str(estimator)
 		else:
-			raise ValueError("Dump flavour {0} not in {1}".format(dump_flavour, ["dill", "joblib"]))
+			if not _is_supported(estimator):
+				raise TypeError("The estimator object is not an instance of {0}".format(BaseEstimator.__name__))
+
+			if with_repr:
+				estimator.repr_ = repr(estimator)
+
+			# if isinstance(estimator, Pipeline):
+			if hasattr(estimator, "_final_estimator"):
+				final_estimator = estimator._final_estimator
+			else:
+				final_estimator = estimator			
+			# if isinstance(final_estimator, H2OEstimator):
+			if hasattr(final_estimator, "download_mojo"):
+				if dump_flavour != "dill":
+					warnings.warn("Changing dump flavour to dill")
+					dump_flavour = "dill"
+				# Avoid MOJO (re-)download if the indicator attribute is set
+				if not (hasattr(final_estimator, "_mojo_path") or hasattr(final_estimator, "_mojo_bytes")):
+					mojo_path = final_estimator.download_mojo()
+					final_estimator._mojo_path = mojo_path
+					dumps.append(mojo_path)
+
+			if dump_flavour == "dill":
+				pkl_path = _dill_dump(estimator, "estimator")
+			elif dump_flavour == "joblib":
+				pkl_path = _joblib_dump(estimator, "estimator")
+			else:
+				raise ValueError("Dump flavour {0} not in {1}".format(dump_flavour, ["dill", "joblib"]))
+			dumps.append(pkl_path)
+
+		java_args = ["-cp", os.pathsep.join(_classpath(user_classpath)), "com.sklearn2pmml.Main"]
 		java_args.extend(["--pkl-input", pkl_path])
-		dumps.append(pkl_path)
 		java_args.extend(["--pmml-output", pmml_path])
 		if pmml_schema:
 			java_args.extend(["--pmml-schema", pmml_schema])
diff --git a/sklearn2pmml/cli.py b/sklearn2pmml/cli.py
index 7a205c1..efe6e55 100644
--- a/sklearn2pmml/cli.py
+++ b/sklearn2pmml/cli.py
@@ -1,8 +1,6 @@
 from argparse import ArgumentParser
 from sklearn2pmml import __version__, sklearn2pmml
 
-import joblib
-
 def main():
 	version = "SkLearn2PMML {}".format(__version__)
 
@@ -14,6 +12,4 @@ def main():
 
 	args = parser.parse_args()
 
-	estimator = joblib.load(args.input)
-
-	sklearn2pmml(estimator, pmml_path = args.output, pmml_schema = args.schema)
+	sklearn2pmml(args.input, pmml_path = args.output, pmml_schema = args.schema)