Skip to content

S
Serving

PaddlePaddle / Serving
大约 1 年 前同步成功

通知 186
Star 833
Fork 253
S
Serving
提交 6d76ddff 编写于 作者: M MRXLT 提交者： GitHub
浏览文件
操作

Merge branch 'develop' into 0.3.0-doc-fix

上级 71715007 83ad1f4e
隐藏空白更改
内联 并排
Showing with 305 addition and 292 deletion
README.md
浏览文件 @ 6d76ddff
([简体中文](./README_CN.md)|English)
<p align="center"> <p align="center">
<br> <br>
<img src='doc/serving_logo.png' width = "600" height = "130"> <img src='doc/serving_logo.png' width = "600" height = "130">
<br> <br>
<p> <p>
<p align="center"> <p align="center">
<br> <br>
<a href="https://travis-ci.com/PaddlePaddle/Serving"> <a href="https://travis-ci.com/PaddlePaddle/Serving">
...@@ -23,14 +26,6 @@ We consider deploying deep learning inference service online to be a user-facing ...@@ -23,14 +26,6 @@ We consider deploying deep learning inference service online to be a user-facing
<img src="doc/demo.gif" width="700"> <img src="doc/demo.gif" width="700">
</p> </p>
<h2 align="center">Some Key Features</h2>
- Integrate with Paddle training pipeline seamlessly, most paddle models can be deployed **with one line command**.
- **Industrial serving features** supported, such as models management, online loading, online A/B testing etc.
- **Distributed Key-Value indexing** supported which is especially useful for large scale sparse features as model inputs.
- **Highly concurrent and efficient communication** between clients and servers supported.
- **Multiple programming languages** supported on client side, such as Golang, C++ and python.
- **Extensible framework design** which can support model serving beyond Paddle.
<h2 align="center">Installation</h2> <h2 align="center">Installation</h2>
...@@ -60,8 +55,40 @@ If you need install modules compiled with develop branch, please download packag ...@@ -60,8 +55,40 @@ If you need install modules compiled with develop branch, please download packag
Client package support Centos 7 and Ubuntu 18, or you can use HTTP service without install client. Client package support Centos 7 and Ubuntu 18, or you can use HTTP service without install client.
<h2 align="center"> Pre-built services with Paddle Serving</h2>
<h3 align="center">Chinese Word Segmentation</h4>
``` shell
> python -m paddle_serving_app.package -get_model lac
> tar -xzf lac.tar.gz
> python lac_web_service.py 9292 &
> curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"words": "我爱北京天安门"}], "fetch":["word_seg"]}' http://127.0.0.1:9393/lac/prediction
{"result":[{"word_seg":"我|爱|北京|天安门"}]}
```
<h3 align="center">Image Classification</h4>
<p align="center">
<br>
<img src='https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg' width = "200" height = "200">
<br>
<p>
``` shell
> python -m paddle_serving_app.package -get_model resnet_v2_50_imagenet
> tar -xzf resnet_v2_50_imagenet.tar.gz
> python resnet50_imagenet_classify.py resnet50_serving_model &
> curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"image": "https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg"}], "fetch": ["score"]}' http://127.0.0.1:9292/image/prediction
{"result":{"label":["daisy"],"prob":[0.9341403245925903]}}
```
<h2 align="center">Quick Start Example</h2> <h2 align="center">Quick Start Example</h2>
This quick start example is only for users who already have a model to deploy and we prepare a ready-to-deploy model here. If you want to know how to use paddle serving from offline training to online serving, please reference to [Train_To_Service](https://github.com/PaddlePaddle/Serving/blob/develop/doc/TRAIN_TO_SERVICE.md)
### Boston House Price Prediction model ### Boston House Price Prediction model
``` shell ``` shell
wget --no-check-certificate https://paddle-serving.bj.bcebos.com/uci_housing.tar.gz wget --no-check-certificate https://paddle-serving.bj.bcebos.com/uci_housing.tar.gz
...@@ -117,138 +144,14 @@ print(fetch_map) ...@@ -117,138 +144,14 @@ print(fetch_map)
``` ```
Here, `client.predict` function has two arguments. `feed` is a `python dict` with model input variable alias name and values. `fetch` assigns the prediction variables to be returned from servers. In the example, the name of `"x"` and `"price"` are assigned when the servable model is saved during training. Here, `client.predict` function has two arguments. `feed` is a `python dict` with model input variable alias name and values. `fetch` assigns the prediction variables to be returned from servers. In the example, the name of `"x"` and `"price"` are assigned when the servable model is saved during training.
<h2 align="center"> Pre-built services with Paddle Serving</h2> <h2 align="center">Some Key Features of Paddle Serving</h2>
<h3 align="center">Chinese Word Segmentation</h4>
- **Description**:
``` shell
Chinese word segmentation HTTP service that can be deployed with one line command.
```
- **Download Servable Package**:
``` shell
wget --no-check-certificate https://paddle-serving.bj.bcebos.com/lac/lac_model_jieba_web.tar.gz
```
- **Host web service**:
``` shell
tar -xzf lac_model_jieba_web.tar.gz
python lac_web_service.py jieba_server_model/ lac_workdir 9292
```
- **Request sample**:
``` shell
curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"words": "我爱北京天安门"}], "fetch":["word_seg"]}' http://127.0.0.1:9292/lac/prediction
```
- **Request result**:
``` shell
{"word_seg":"我|爱|北京|天安门"}
```
<h3 align="center">Image Classification</h4>
- **Description**:
``` shell
Image classification trained with Imagenet dataset. A label and corresponding probability will be returned.
Note: This demo needs paddle-serving-server-gpu.
```
- **Download Servable Package**:
``` shell
wget --no-check-certificate https://paddle-serving.bj.bcebos.com/imagenet-example/imagenet_demo.tar.gz
```
- **Host web service**:
``` shell
tar -xzf imagenet_demo.tar.gz
python image_classification_service_demo.py resnet50_serving_model
```
- **Request sample**:
<p align="center">
<br>
<img src='https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg' width = "200" height = "200">
<br>
<p>
``` shell
curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"url": "https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg"}], "fetch": ["score"]}' http://127.0.0.1:9292/image/prediction
```
- **Request result**:
``` shell
{"label":"daisy","prob":0.9341403245925903}
```
<h3 align="center">More Demos</h3>
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | Bert-Base-Baike |
| URL | [https://paddle-serving.bj.bcebos.com/bert_example/bert_seq128.tar.gz](https://paddle-serving.bj.bcebos.com/bert_example%2Fbert_seq128.tar.gz) |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/bert |
| Description | Get semantic representation from a Chinese Sentence |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | Resnet50-Imagenet |
| URL | [https://paddle-serving.bj.bcebos.com/imagenet-example/ResNet50_vd.tar.gz](https://paddle-serving.bj.bcebos.com/imagenet-example%2FResNet50_vd.tar.gz) |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imagenet |
| Description | Get image semantic representation from an image |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | Resnet101-Imagenet |
| URL | https://paddle-serving.bj.bcebos.com/imagenet-example/ResNet101_vd.tar.gz |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imagenet |
| Description | Get image semantic representation from an image |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | CNN-IMDB |
| URL | https://paddle-serving.bj.bcebos.com/imdb-demo/imdb_model.tar.gz |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imdb |
| Description | Get category probability from an English Sentence |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | LSTM-IMDB |
| URL | https://paddle-serving.bj.bcebos.com/imdb-demo/imdb_model.tar.gz |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imdb |
| Description | Get category probability from an English Sentence |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | BOW-IMDB |
| URL | https://paddle-serving.bj.bcebos.com/imdb-demo/imdb_model.tar.gz |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imdb |
| Description | Get category probability from an English Sentence |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | Jieba-LAC |
| URL | https://paddle-serving.bj.bcebos.com/lac/lac_model.tar.gz |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/lac |
| Description | Get word segmentation from a Chinese Sentence |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| Model Name | DNN-CTR |
| URL | https://paddle-serving.bj.bcebos.com/criteo_ctr_example/criteo_ctr_demo_model.tar.gz |
| Client/Server Code | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/criteo_ctr |
| Description | Get click probability from a feature vector of item |
- Integrate with Paddle training pipeline seamlessly, most paddle models can be deployed **with one line command**.
- **Industrial serving features** supported, such as models management, online loading, online A/B testing etc.
- **Distributed Key-Value indexing** supported which is especially useful for large scale sparse features as model inputs.
- **Highly concurrent and efficient communication** between clients and servers supported.
- **Multiple programming languages** supported on client side, such as Golang, C++ and python.
- **Extensible framework design** which can support model serving beyond Paddle.
<h2 align="center">Document</h2> <h2 align="center">Document</h2>
......
README_CN.md
浏览文件 @ 6d76ddff
(简体中文|[English](./README.md))
<p align="center"> <p align="center">
<br> <br>
<img src='https://paddle-serving.bj.bcebos.com/imdb-demo%2FLogoMakr-3Bd2NM-300dpi.png' width = "600" height = "130"> <img src='https://paddle-serving.bj.bcebos.com/imdb-demo%2FLogoMakr-3Bd2NM-300dpi.png' width = "600" height = "130">
<br> <br>
<p> <p>
<p align="center"> <p align="center">
<br> <br>
<a href="https://travis-ci.com/PaddlePaddle/Serving"> <a href="https://travis-ci.com/PaddlePaddle/Serving">
...@@ -24,14 +27,7 @@ Paddle Serving 旨在帮助深度学习开发者轻易部署在线预测服务 ...@@ -24,14 +27,7 @@ Paddle Serving 旨在帮助深度学习开发者轻易部署在线预测服务
<img src="doc/demo.gif" width="700"> <img src="doc/demo.gif" width="700">
</p> </p>
<h2 align="center">核心功能</h2>
- 与Paddle训练紧密连接，绝大部分Paddle模型可以 **一键部署**.
- 支持 **工业级的服务能力** 例如模型管理，在线加载，在线A/B测试等.
- 支持 **分布式键值对索引** 助力于大规模稀疏特征作为模型输入.
- 支持客户端和服务端之间 **高并发和高效通信**.
- 支持 **多种编程语言** 开发客户端，例如Golang，C++和Python.
- **可伸缩框架设计** 可支持不限于Paddle的模型服务.
<h2 align="center">安装</h2> <h2 align="center">安装</h2>
...@@ -61,7 +57,38 @@ pip install paddle-serving-server-gpu # GPU ...@@ -61,7 +57,38 @@ pip install paddle-serving-server-gpu # GPU
客户端安装包支持Centos 7和Ubuntu 18，或者您可以使用HTTP服务，这种情况下不需要安装客户端。 客户端安装包支持Centos 7和Ubuntu 18，或者您可以使用HTTP服务，这种情况下不需要安装客户端。
<h2 align="center">快速启动示例</h2> <h2 align="center"> Paddle Serving预装的服务 </h2>
<h3 align="center">中文分词</h4>
``` shell
> python -m paddle_serving_app.package -get_model lac
> tar -xzf lac.tar.gz
> python lac_web_service.py 9292 &
> curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"words": "我爱北京天安门"}], "fetch":["word_seg"]}' http://127.0.0.1:9393/lac/prediction
{"result":[{"word_seg":"我|爱|北京|天安门"}]}
```
<h3 align="center">图像分类</h4>
<p align="center">
<br>
<img src='https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg' width = "200" height = "200">
<br>
<p>
``` shell
> python -m paddle_serving_app.package -get_model resnet_v2_50_imagenet
> tar -xzf resnet_v2_50_imagenet.tar.gz
> python resnet50_imagenet_classify.py resnet50_serving_model &
> curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"image": "https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg"}], "fetch": ["score"]}' http://127.0.0.1:9292/image/prediction
{"result":{"label":["daisy"],"prob":[0.9341403245925903]}}
```
<h2 align="center">快速开始示例</h2>
这个快速开始示例主要是为了给那些已经有一个要部署的模型的用户准备的，而且我们也提供了一个可以用来部署的模型。如果您想知道如何从离线训练到在线服务走完全流程，请参考[从训练到部署](https://github.com/PaddlePaddle/Serving/blob/develop/doc/TRAIN_TO_SERVICE_CN.md)
<h3 align="center">波士顿房价预测</h3> <h3 align="center">波士顿房价预测</h3>
...@@ -122,139 +149,14 @@ print(fetch_map) ...@@ -122,139 +149,14 @@ print(fetch_map)
``` ```
在这里，`client.predict`函数具有两个参数。 `feed`是带有模型输入变量别名和值的`python dict`。 `fetch`被要从服务器返回的预测变量赋值。 在该示例中，在训练过程中保存可服务模型时，被赋值的tensor名为`"x"`和`"price"`。 在这里，`client.predict`函数具有两个参数。 `feed`是带有模型输入变量别名和值的`python dict`。 `fetch`被要从服务器返回的预测变量赋值。 在该示例中，在训练过程中保存可服务模型时，被赋值的tensor名为`"x"`和`"price"`。
<h2 align="center">Paddle Serving预装的服务</h2> <h2 align="center">Paddle Serving的核心功能</h2>
<h3 align="center">中文分词模型</h4>
- **介绍**:
``` shell
本示例为中文分词HTTP服务一键部署
```
- **下载服务包**:
``` shell
wget --no-check-certificate https://paddle-serving.bj.bcebos.com/lac/lac_model_jieba_web.tar.gz
```
- **启动web服务**:
``` shell
tar -xzf lac_model_jieba_web.tar.gz
python lac_web_service.py jieba_server_model/ lac_workdir 9292
```
- **客户端请求示例**:
``` shell
curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"words": "我爱北京天安门"}], "fetch":["word_seg"]}' http://127.0.0.1:9292/lac/prediction
```
- **返回结果示例**:
``` shell
{"word_seg":"我|爱|北京|天安门"}
```
<h3 align="center">图像分类模型</h4>
- **介绍**:
``` shell
图像分类模型由Imagenet数据集训练而成，该服务会返回一个标签及其概率
注意：本示例需要安装paddle-serving-server-gpu
```
- **下载服务包**:
``` shell
wget --no-check-certificate https://paddle-serving.bj.bcebos.com/imagenet-example/imagenet_demo.tar.gz
```
- **启动web服务**:
``` shell
tar -xzf imagenet_demo.tar.gz
python image_classification_service_demo.py resnet50_serving_model
```
- **客户端请求示例**:
<p align="center">
<br>
<img src='https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg' width = "200" height = "200">
<br>
<p>
``` shell
curl -H "Content-Type:application/json" -X POST -d '{"feed":[{"url": "https://paddle-serving.bj.bcebos.com/imagenet-example/daisy.jpg"}], "fetch": ["score"]}' http://127.0.0.1:9292/image/prediction
```
- **返回结果示例**:
``` shell
{"label":"daisy","prob":0.9341403245925903}
```
<h3 align="center">更多示例</h3>
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | Bert-Base-Baike |
| 下载链接 | [https://paddle-serving.bj.bcebos.com/bert_example/bert_seq128.tar.gz](https://paddle-serving.bj.bcebos.com/bert_example%2Fbert_seq128.tar.gz) |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/bert |
| 介绍 | 获得一个中文语句的语义表示 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | Resnet50-Imagenet |
| 下载链接 | [https://paddle-serving.bj.bcebos.com/imagenet-example/ResNet50_vd.tar.gz](https://paddle-serving.bj.bcebos.com/imagenet-example%2FResNet50_vd.tar.gz) |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imagenet |
| 介绍 | 获得一张图片的图像语义表示 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | Resnet101-Imagenet |
| 下载链接 | https://paddle-serving.bj.bcebos.com/imagenet-example/ResNet101_vd.tar.gz |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imagenet |
| 介绍 | 获得一张图片的图像语义表示 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | CNN-IMDB |
| 下载链接 | https://paddle-serving.bj.bcebos.com/imdb-demo/imdb_model.tar.gz |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imdb |
| 介绍 | 从一个中文语句获得类别及其概率 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | LSTM-IMDB |
| 下载链接 | https://paddle-serving.bj.bcebos.com/imdb-demo/imdb_model.tar.gz |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imdb |
| 介绍 | 从一个英文语句获得类别及其概率 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | BOW-IMDB |
| 下载链接 | https://paddle-serving.bj.bcebos.com/imdb-demo/imdb_model.tar.gz |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/imdb |
| 介绍 | 从一个英文语句获得类别及其概率 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | Jieba-LAC |
| 下载链接 | https://paddle-serving.bj.bcebos.com/lac/lac_model.tar.gz |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/lac |
| 介绍 | 获取中文语句的分词 |
| Key | Value |
| :----------------- | :----------------------------------------------------------- |
| 模型名 | DNN-CTR |
| 下载链接 | https://paddle-serving.bj.bcebos.com/criteo_ctr_example/criteo_ctr_demo_model.tar.gz |
| 客户端/服务端代码 | https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/criteo_ctr |
| 介绍 | 从项目的特征向量中获得点击概率 |
- 与Paddle训练紧密连接，绝大部分Paddle模型可以 **一键部署**.
- 支持 **工业级的服务能力** 例如模型管理，在线加载，在线A/B测试等.
- 支持 **分布式键值对索引** 助力于大规模稀疏特征作为模型输入.
- 支持客户端和服务端之间 **高并发和高效通信**.
- 支持 **多种编程语言** 开发客户端，例如Golang，C++和Python.
- **可伸缩框架设计** 可支持不限于Paddle的模型服务.
<h2 align="center">文档</h2> <h2 align="center">文档</h2>
......
python/examples/bert/bert_web_service.py
浏览文件 @ 6d76ddff
...@@ -21,7 +21,10 @@ import os ...@@ -21,7 +21,10 @@ import os
class BertService(WebService): class BertService(WebService):
def load(self): def load(self):
self.reader = ChineseBertReader(vocab_file="vocab.txt", max_seq_len=128) self.reader = ChineseBertReader({
"vocab_file": "vocab.txt",
"max_seq_len": 128
})
def preprocess(self, feed=[], fetch=[]): def preprocess(self, feed=[], fetch=[]):
feed_res = [ feed_res = [
......
python/examples/deeplabv3/README.md 0 → 100644
浏览文件 @ 6d76ddff
# Image Segmentation
## Get Model
```
python -m paddle_serving_app.package --get_model deeplabv3
tar -xzvf deeplabv3.tar.gz
```
## RPC Service
### Start Service
```
python -m paddle_serving_server_gpu.serve --model deeplabv3_server --gpu_ids 0 --port 9494
```
### Client Prediction
```
python deeplabv3_client.py
```
python/examples/deeplabv3/README_CN.md 0 → 100644
浏览文件 @ 6d76ddff
# 图像分割
## 获取模型
```
python -m paddle_serving_app.package --get_model deeplabv3
tar -xzvf deeplabv3.tar.gz
```
## RPC 服务
### 启动服务端
```
python -m paddle_serving_server_gpu.serve --model deeplabv3_server --gpu_ids 0 --port 9494
```
### 客户端预测
```
python deeplabv3_client.py
python/examples/deeplabv3/deeplabv3_client.py
浏览文件 @ 6d76ddff
...@@ -18,7 +18,7 @@ import sys ...@@ -18,7 +18,7 @@ import sys
import cv2 import cv2
client = Client() client = Client()
client.load_client_config("seg_client/serving_client_conf.prototxt") client.load_client_config("deeplabv3_client/serving_client_conf.prototxt")
client.connect(["127.0.0.1:9494"]) client.connect(["127.0.0.1:9494"])
preprocess = Sequential( preprocess = Sequential(
......
python/examples/faster_rcnn_model/README.md
浏览文件 @ 6d76ddff
...@@ -12,8 +12,8 @@ If you want to have more detection models, please refer to [Paddle Detection Mod ...@@ -12,8 +12,8 @@ If you want to have more detection models, please refer to [Paddle Detection Mod
### Start the service ### Start the service
``` ```
tar xf faster_rcnn_model.tar.gz tar xf faster_rcnn_model.tar.gz
mv faster_rcnn_model/pddet *. mv faster_rcnn_model/pddet* .
GLOG_v=2 python -m paddle_serving_server_gpu.serve --model pddet_serving_model --port 9494 --gpu_id 0 GLOG_v=2 python -m paddle_serving_server_gpu.serve --model pddet_serving_model --port 9494 --gpu_ids 0
``` ```
### Perform prediction ### Perform prediction
......
python/examples/faster_rcnn_model/README_CN.md
浏览文件 @ 6d76ddff
...@@ -13,7 +13,7 @@ wget https://paddle-serving.bj.bcebos.com/pddet_demo/infer_cfg.yml ...@@ -13,7 +13,7 @@ wget https://paddle-serving.bj.bcebos.com/pddet_demo/infer_cfg.yml
``` ```
tar xf faster_rcnn_model.tar.gz tar xf faster_rcnn_model.tar.gz
mv faster_rcnn_model/pddet* ./ mv faster_rcnn_model/pddet* ./
GLOG_v=2 python -m paddle_serving_server_gpu.serve --model pddet_serving_model --port 9494 --gpu_id 0 GLOG_v=2 python -m paddle_serving_server_gpu.serve --model pddet_serving_model --port 9494 --gpu_ids 0
``` ```
### 执行预测 ### 执行预测
......
python/examples/imagenet/README_CN.md
浏览文件 @ 6d76ddff
...@@ -19,10 +19,10 @@ pip install paddle_serving_app ...@@ -19,10 +19,10 @@ pip install paddle_serving_app
启动server端 启动server端
``` ```
python image_classification_service.py ResNet50_vd_model cpu 9696 #cpu预测服务 python resnet50_web_service.py ResNet50_vd_model cpu 9696 #cpu预测服务
``` ```
``` ```
python image_classification_service.py ResNet50_vd_model gpu 9696 #gpu预测服务 python resnet50_web_service.py ResNet50_vd_model gpu 9696 #gpu预测服务
``` ```
......
python/examples/mobilenet/README.md 0 → 100644
浏览文件 @ 6d76ddff
# Image Classification
## Get Model
```
python -m paddle_serving_app.package --get_model mobilenet_v2_imagenet
tar -xzvf mobilenet_v2_imagenet.tar.gz
```
## RPC Service
### Start Service
```
python -m paddle_serving_server_gpu.serve --model mobilenet_v2_imagenet_model --gpu_ids 0 --port 9393
```
### Client Prediction
```
python mobilenet_tutorial.py
```
python/examples/mobilenet/README_CN.md 0 → 100644
浏览文件 @ 6d76ddff
# 图像分类
## 获取模型
```
python -m paddle_serving_app.package --get_model mobilenet_v2_imagenet
tar -xzvf mobilenet_v2_imagenet.tar.gz
```
## RPC 服务
### 启动服务端
```
python -m paddle_serving_server_gpu.serve --model mobilenet_v2_imagenet_model --gpu_ids 0 --port 9393
```
### 客户端预测
```
python mobilenet_tutorial.py
```
python/examples/resnet_v2_50/README.md 0 → 100644
浏览文件 @ 6d76ddff
# Image Classification
## Get Model
```
python -m paddle_serving_app.package --get_model resnet_v2_50_imagenet
tar -xzvf resnet_v2_50_imagenet.tar.gz
```
## RPC Service
### Start Service
```
python -m paddle_serving_server_gpu.serve --model resnet_v2_50_imagenet_model --gpu_ids 0 --port 9393
```
### Client Prediction
```
python resnet50_v2_tutorial.py
```
python/examples/resnet_v2_50/README_CN.md 0 → 100644
浏览文件 @ 6d76ddff
# 图像分类
## 获取模型
```
python -m paddle_serving_app.package --get_model resnet_v2_50_imagenet
tar -xzvf resnet_v2_50_imagenet.tar.gz
```
## RPC 服务
### 启动服务端
```
python -m paddle_serving_server_gpu.serve --model resnet_v2_50_imagenet_model --gpu_ids 0 --port 9393
```
### 客户端预测
```
python resnet50_v2_tutorial.py
```
python/examples/unet_for_image_seg/README.md 0 → 100644
浏览文件 @ 6d76ddff
# Image Segmentation
## Get Model
```
python -m paddle_serving_app.package --get_model unet
tar -xzvf unet.tar.gz
```
## RPC Service
### Start Service
```
python -m paddle_serving_server_gpu.serve --model unet_model --gpu_ids 0 --port 9494
```
### Client Prediction
```
python seg_client.py
```
python/examples/unet_for_image_seg/README_CN.md 0 → 100644
浏览文件 @ 6d76ddff
# 图像分割
## 获取模型
```
python -m paddle_serving_app.package --get_model unet
tar -xzvf unet.tar.gz
```
## RPC 服务
### 启动服务端
```
python -m paddle_serving_server_gpu.serve --model unet_model --gpu_ids 0 --port 9494
```
### 客户端预测
```
python seg_client.py
```
python/paddle_serving_app/README.md
浏览文件 @ 6d76ddff
...@@ -12,7 +12,7 @@ pip install paddle_serving_app ...@@ -12,7 +12,7 @@ pip install paddle_serving_app
## Get model list ## Get model list
```shell ```shell
python -m paddle_serving_app.package --model_list python -m paddle_serving_app.package --list_model
``` ```
## Download pre-training model ## Download pre-training model
......
python/paddle_serving_app/README_CN.md
浏览文件 @ 6d76ddff
...@@ -11,7 +11,7 @@ pip install paddle_serving_app ...@@ -11,7 +11,7 @@ pip install paddle_serving_app
## 获取模型列表 ## 获取模型列表
```shell ```shell
python -m paddle_serving_app.package --model_list python -m paddle_serving_app.package --list_model
``` ```
## 下载预训练模型 ## 下载预训练模型
......
python/paddle_serving_app/reader/image_reader.py
浏览文件 @ 6d76ddff
...@@ -296,7 +296,10 @@ class File2Image(object): ...@@ -296,7 +296,10 @@ class File2Image(object):
pass pass
def __call__(self, img_path): def __call__(self, img_path):
fin = open(img_path) if py_version == 2:
fin = open(img_path)
else:
fin = open(img_path, "rb")
sample = fin.read() sample = fin.read()
data = np.fromstring(sample, np.uint8) data = np.fromstring(sample, np.uint8)
img = cv2.imdecode(data, cv2.IMREAD_COLOR) img = cv2.imdecode(data, cv2.IMREAD_COLOR)
......
python/paddle_serving_client/__init__.py
浏览文件 @ 6d76ddff
...@@ -61,13 +61,18 @@ class SDKConfig(object): ...@@ -61,13 +61,18 @@ class SDKConfig(object):
self.tag_list = [] self.tag_list = []
self.cluster_list = [] self.cluster_list = []
self.variant_weight_list = [] self.variant_weight_list = []
self.rpc_timeout_ms = 20000
self.load_balance_strategy = "la"
def add_server_variant(self, tag, cluster, variant_weight): def add_server_variant(self, tag, cluster, variant_weight):
self.tag_list.append(tag) self.tag_list.append(tag)
self.cluster_list.append(cluster) self.cluster_list.append(cluster)
self.variant_weight_list.append(variant_weight) self.variant_weight_list.append(variant_weight)
def gen_desc(self): def set_load_banlance_strategy(self, strategy):
self.load_balance_strategy = strategy
def gen_desc(self, rpc_timeout_ms):
predictor_desc = sdk.Predictor() predictor_desc = sdk.Predictor()
predictor_desc.name = "general_model" predictor_desc.name = "general_model"
predictor_desc.service_name = \ predictor_desc.service_name = \
...@@ -86,7 +91,7 @@ class SDKConfig(object): ...@@ -86,7 +91,7 @@ class SDKConfig(object):
self.sdk_desc.predictors.extend([predictor_desc]) self.sdk_desc.predictors.extend([predictor_desc])
self.sdk_desc.default_variant_conf.tag = "default" self.sdk_desc.default_variant_conf.tag = "default"
self.sdk_desc.default_variant_conf.connection_conf.connect_timeout_ms = 2000 self.sdk_desc.default_variant_conf.connection_conf.connect_timeout_ms = 2000
self.sdk_desc.default_variant_conf.connection_conf.rpc_timeout_ms = 20000 self.sdk_desc.default_variant_conf.connection_conf.rpc_timeout_ms = rpc_timeout_ms
self.sdk_desc.default_variant_conf.connection_conf.connect_retry_count = 2 self.sdk_desc.default_variant_conf.connection_conf.connect_retry_count = 2
self.sdk_desc.default_variant_conf.connection_conf.max_connection_per_host = 100 self.sdk_desc.default_variant_conf.connection_conf.max_connection_per_host = 100
self.sdk_desc.default_variant_conf.connection_conf.hedge_request_timeout_ms = -1 self.sdk_desc.default_variant_conf.connection_conf.hedge_request_timeout_ms = -1
...@@ -119,6 +124,7 @@ class Client(object): ...@@ -119,6 +124,7 @@ class Client(object):
self.profile_ = _Profiler() self.profile_ = _Profiler()
self.all_numpy_input = True self.all_numpy_input = True
self.has_numpy_input = False self.has_numpy_input = False
self.rpc_timeout_ms = 20000
def load_client_config(self, path): def load_client_config(self, path):
from .serving_client import PredictorClient from .serving_client import PredictorClient
...@@ -171,6 +177,12 @@ class Client(object): ...@@ -171,6 +177,12 @@ class Client(object):
self.predictor_sdk_.add_server_variant(tag, cluster, self.predictor_sdk_.add_server_variant(tag, cluster,
str(variant_weight)) str(variant_weight))
def set_rpc_timeout_ms(self, rpc_timeout):
if not isinstance(rpc_timeout, int):
raise ValueError("rpc_timeout must be int type.")
else:
self.rpc_timeout_ms = rpc_timeout
def connect(self, endpoints=None): def connect(self, endpoints=None):
# check whether current endpoint is available # check whether current endpoint is available
# init from client config # init from client config
...@@ -188,7 +200,7 @@ class Client(object): ...@@ -188,7 +200,7 @@ class Client(object):
print( print(
"parameter endpoints({}) will not take effect, because you use the add_variant function.". "parameter endpoints({}) will not take effect, because you use the add_variant function.".
format(endpoints)) format(endpoints))
sdk_desc = self.predictor_sdk_.gen_desc() sdk_desc = self.predictor_sdk_.gen_desc(self.rpc_timeout_ms)
self.client_handle_.create_predictor_by_desc(sdk_desc.SerializeToString( self.client_handle_.create_predictor_by_desc(sdk_desc.SerializeToString(
)) ))
......
python/paddle_serving_server/__init__.py
浏览文件 @ 6d76ddff
...@@ -23,6 +23,7 @@ import paddle_serving_server as paddle_serving_server ...@@ -23,6 +23,7 @@ import paddle_serving_server as paddle_serving_server
from .version import serving_server_version from .version import serving_server_version
from contextlib import closing from contextlib import closing
import collections import collections
import fcntl
class OpMaker(object): class OpMaker(object):
...@@ -322,6 +323,10 @@ class Server(object): ...@@ -322,6 +323,10 @@ class Server(object):
bin_url = "https://paddle-serving.bj.bcebos.com/bin/" + tar_name bin_url = "https://paddle-serving.bj.bcebos.com/bin/" + tar_name
self.server_path = os.path.join(self.module_path, floder_name) self.server_path = os.path.join(self.module_path, floder_name)
#acquire lock
version_file = open("{}/version.py".format(self.module_path), "r")
fcntl.flock(version_file, fcntl.LOCK_EX)
if not os.path.exists(self.server_path): if not os.path.exists(self.server_path):
print('Frist time run, downloading PaddleServing components ...') print('Frist time run, downloading PaddleServing components ...')
r = os.system('wget ' + bin_url + ' --no-check-certificate') r = os.system('wget ' + bin_url + ' --no-check-certificate')
...@@ -345,6 +350,8 @@ class Server(object): ...@@ -345,6 +350,8 @@ class Server(object):
foemat(self.module_path)) foemat(self.module_path))
finally: finally:
os.remove(tar_name) os.remove(tar_name)
#release lock
version_file.close()
os.chdir(self.cur_path) os.chdir(self.cur_path)
self.bin_path = self.server_path + "/serving" self.bin_path = self.server_path + "/serving"
......
python/paddle_serving_server_gpu/__init__.py
浏览文件 @ 6d76ddff
...@@ -25,6 +25,7 @@ from .version import serving_server_version ...@@ -25,6 +25,7 @@ from .version import serving_server_version
from contextlib import closing from contextlib import closing
import argparse import argparse
import collections import collections
import fcntl
def serve_args(): def serve_args():
...@@ -347,6 +348,11 @@ class Server(object): ...@@ -347,6 +348,11 @@ class Server(object):
download_flag = "{}/{}.is_download".format(self.module_path, download_flag = "{}/{}.is_download".format(self.module_path,
folder_name) folder_name)
#acquire lock
version_file = open("{}/version.py".format(self.module_path), "r")
fcntl.flock(version_file, fcntl.LOCK_EX)
if os.path.exists(download_flag): if os.path.exists(download_flag):
os.chdir(self.cur_path) os.chdir(self.cur_path)
self.bin_path = self.server_path + "/serving" self.bin_path = self.server_path + "/serving"
...@@ -377,6 +383,8 @@ class Server(object): ...@@ -377,6 +383,8 @@ class Server(object):
format(self.module_path)) format(self.module_path))
finally: finally:
os.remove(tar_name) os.remove(tar_name)
#release lock
version_file.cloes()
os.chdir(self.cur_path) os.chdir(self.cur_path)
self.bin_path = self.server_path + "/serving" self.bin_path = self.server_path + "/serving"
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或