Merge pull request #268 from guru4elephant/fix_web_service_bug

add gpu web service scripts

python/examples/imagenet/get_model.sh

-wget https://paddle-serving.bj.bcebos.com/imagenet-example/conf_and_model.tar.gz
+wget --no-check-certificate https://paddle-serving.bj.bcebos.com/imagenet-example/conf_and_model.tar.gz
 tar -xzvf conf_and_model.tar.gz

python/examples/imagenet/image_classification_service_gpu.py

+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from paddle_serving_server_gpu.web_service import WebService
+import sys
+import os
+import base64
+from image_reader import ImageReader
+
+
+class ImageService(WebService):
+    """
+    preprocessing function for image classification
+    """
+
+    def preprocess(self, feed={}, fetch=[]):
+        reader = ImageReader()
+        if "image" not in feed:
+            raise ("feed data error!")
+        sample = base64.b64decode(feed["image"])
+        img = reader.process_image(sample)
+        res_feed = {}
+        res_feed["image"] = img.reshape(-1)
+        return res_feed, fetch
+
+
+image_service = ImageService(name="image")
+image_service.load_model_config(sys.argv[1])
+gpu_ids = os.environ["CUDA_VISIBLE_DEVICES"]
+gpus = [int(x) for x in gpu_ids.split(",")]
+image_service.set_gpus(gpus)
+image_service.prepare_server(
+    workdir=sys.argv[2], port=int(sys.argv[3]), device="gpu")
+image_service.run_server()

python/examples/imagenet/image_http_client.py

@@ -15,6 +15,7 @@
 import requests
 import base64
 import json
+import time
 
 
 def predict(image_path, server):
@@ -22,10 +23,14 @@ def predict(image_path, server):
     req = json.dumps({"image": image, "fetch": ["score"]})
     r = requests.post(
         server, data=req, headers={"Content-Type": "application/json"})
-    print(r.json()["score"])
 
 
 if __name__ == "__main__":
-    server = "http://127.0.0.1:9393/image/prediction"
+    server = "http://127.0.0.1:9292/image/prediction"
     image_path = "./data/n01440764_10026.JPEG"
-    predict(image_path, server)
+    start = time.time()
+    for i in range(1000):
+        predict(image_path, server)
+        print(i)
+    end = time.time()
+    print(end - start)

python/examples/imagenet/image_rpc_client.py

@@ -15,15 +15,21 @@
 import sys
 from image_reader import ImageReader
 from paddle_serving_client import Client
+import time
 
 client = Client()
 client.load_client_config(sys.argv[1])
-client.connect(["127.0.0.1:9393"])
+client.connect(["127.0.0.1:9292"])
 reader = ImageReader()
-with open("./data/n01440764_10026.JPEG") as f:
-    img = f.read()
 
-img = reader.process_image(img).reshape(-1)
-fetch_map = client.predict(feed={"image": img}, fetch=["score"])
+start = time.time()
+for i in range(1000):
+    with open("./data/n01440764_10026.JPEG") as f:
+        img = f.read()
+    img = reader.process_image(img).reshape(-1)
+    fetch_map = client.predict(feed={"image": img}, fetch=["score"])
+    print(i)
+end = time.time()
+print(end - start)
 
-print(fetch_map["score"])
+#print(fetch_map["score"])

python/paddle_serving_server_gpu/serve.py

@@ -64,6 +64,7 @@ def start_gpu_card_model(gpuid, args):  # pylint: disable=doc-string-missing
 def start_multi_card(args):  # pylint: disable=doc-string-missing
     gpus = ""
     if args.gpu_ids == "":
+        import os
         gpus = os.environ["CUDA_VISIBLE_DEVICES"]
     else:
         gpus = args.gpu_ids.split(",")

python/paddle_serving_server_gpu/web_serve.py

@@ -36,4 +36,4 @@ if __name__ == "__main__":
         web_service.set_gpus(gpus)
     web_service.prepare_server(
         workdir=args.workdir, port=args.port, device=args.device)
-    service.run_server()
+    web_service.run_server()

python/paddle_serving_server_gpu/web_service.py

@@ -107,6 +107,8 @@ class WebService(object):
             if "fetch" not in request.json:
                 abort(400)
             feed, fetch = self.preprocess(request.json, request.json["fetch"])
+            if "fetch" in feed:
+                del feed["fetch"]
             fetch_map = client_list[0].predict(feed=feed, fetch=fetch)
             fetch_map = self.postprocess(
                 feed=request.json, fetch=fetch, fetch_map=fetch_map)