ResNet Latency test

examples/models/resnet/Makefile

@@ -0,0 +1,12 @@
+
+
+build_image:
+	s2i build -E environment_grpc . seldonio/seldon-core-s2i-python36:0.4-SNAPSHOT seldon-resnet2.4
+
+
+clean:
+	rm -rf model
+	rm -rf proto/__pycache__
+	rm -f proto/*.py
+	rm -r proto/*.proto
+	rm -rf tensorflow

examples/models/resnet/Resnet.py

@@ -0,0 +1,38 @@
+import tensorflow as tf
+import numpy as np
+import logging
+import datetime
+
+def get_logger(name):
+    logger = logging.getLogger(name)
+    log_formatter = logging.Formatter("%(asctime)s - %(name)s - "
+                                      "%(levelname)s - %(message)s")
+    logger.setLevel('DEBUG')
+
+    console_handler = logging.StreamHandler()
+    console_handler.setFormatter(log_formatter)
+    logger.addHandler(console_handler)
+
+    return logger
+
+logger = get_logger(__name__)
+
+class Resnet(object):
+    def __init__(self):
+        self.class_names = ["class:{}".format(str(i)) for i in range(1000)]
+        self.sess = tf.Session()
+        tf.saved_model.loader.load(self.sess, ["serve"], "model", clear_devices=True)
+
+        graph = tf.get_default_graph()
+        self.x = graph.get_tensor_by_name("input:0")
+        self.y = graph.get_tensor_by_name("resnet_v1_50/predictions/Reshape_1:0")
+
+    def predict(self,X,feature_names):
+        start_time = datetime.datetime.now()
+        predictions = self.sess.run(self.y,feed_dict={self.x:X})
+        end_time = datetime.datetime.now()
+        duration = (end_time - start_time).total_seconds() * 1000
+        logger.debug("Processing time: {:.2f} ms".format(duration))
+        return predictions.astype(np.float64)
+
+

examples/models/resnet/environment_grpc

@@ -0,0 +1,4 @@
+MODEL_NAME=Resnet
+API_TYPE=GRPC
+SERVICE_TYPE=MODEL
+PERSISTENCE=0

examples/models/resnet/requirements.txt

@@ -0,0 +1 @@
+tensorflow