Update sample model to mobilenetv2-int8-tf-0001

This model could show AVX512 VNNI performance advantage.

Signed-off-by: Qi Zheng <qi.zheng@intel.com>
This commit is contained in:
Qi Zheng
2019-12-24 15:51:43 +08:00
committed by qzheng527
parent 936efb6b85
commit fe3c1fb01f
5 changed files with 7 additions and 10 deletions
+4 -5
View File
@@ -59,7 +59,7 @@ The model precision to be chosen, FP32, FP16 or INT8.
#### MODEL_PATH
The chosen model path, automatically set by the entrypoint script.
For example, if using alexnet model to do classification, two environment variables
For example, if using mobilenetv2-int8-tf-0001 model to do classification, two environment variables
need to be passed to the container. Details can refer to the deployment below.
@@ -75,7 +75,7 @@ Docker Hub.
docker pull clearlinux/openvino
```
2. Start one-time classification_sample with alexnet model as below:
2. Start one-time classification_sample with mobilenetv2-int8-tf-0001 model as below:
* Use docker-compose to start the example:
```
@@ -87,7 +87,7 @@ Docker Hub.
Or
3. Start a simple openvino-server to accept image to do classification_sample with alexnet model:
3. Start a simple openvino-server to accept image to do classification_sample with mobilenetv2-int8-tf-0001 model:
* Use docker-compose to start the server first:
```
@@ -123,8 +123,7 @@ To deploy the image on a Kubernetes cluster:
```
kubectl get pods -o wide
```
This may take some time because it requires downloading/converting the
alexnet model.
This may take some time because it requires downloading/converting the model.
Note, if your cluster is behind some proxy, you may need set the proxy
environment in the yaml file to make the model-init can download the model.
+1 -1
View File
@@ -52,7 +52,7 @@ spec:
# - name: https_proxy
# value: <your proxy>
- name: MODEL_NAME
value: alexnet
value: mobilenetv2-int8-tf-0001
- name: MODEL_PRECISION
value: FP32
args:
-2
View File
@@ -2,6 +2,4 @@
# Copyright (C) 2018 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
# demo to use alexnet modle to do classification
classification_sample_async -i cat.bmp -m $MODEL_PATH/$MODEL_NAME.xml
+1 -1
View File
@@ -6,7 +6,7 @@ services:
environment:
http_proxy: $http_proxy
https_proxy: $https_proxy
MODEL_NAME: alexnet
MODEL_NAME: mobilenetv2-int8-tf-0001
MODEL_PRECISION: FP32
ports:
- "5000:5000"
+1 -1
View File
@@ -6,7 +6,7 @@ services:
environment:
http_proxy: $http_proxy
https_proxy: $https_proxy
MODEL_NAME: alexnet
MODEL_NAME: mobilenetv2-int8-tf-0001
MODEL_PRECISION: FP32
volumes:
- "./cat.bmp:/app/cat.bmp"