From 949d964a690549ae9e3c2a441781ea033da53493 Mon Sep 17 00:00:00 2001
From: yy0131 <1595992831@qq.com>
Date: Fri, 23 Sep 2022 10:42:33 +0800
Subject: [PATCH] =?UTF-8?q?add=20codes=E3=80=91?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../cv/resnet50_for_TensorFlow/README.md      | 73 +++++++++++++++++++
 1 file changed, 73 insertions(+)
 create mode 100644 TensorFlow/contrib/cv/resnet50_for_TensorFlow/README.md

diff --git a/TensorFlow/contrib/cv/resnet50_for_TensorFlow/README.md b/TensorFlow/contrib/cv/resnet50_for_TensorFlow/README.md
new file mode 100644
index 000000000..e10a40a05
--- /dev/null
+++ b/TensorFlow/contrib/cv/resnet50_for_TensorFlow/README.md
@@ -0,0 +1,73 @@
+## 模型功能
+
+ 对图像中的物体进行分类。
+
+## 原始模型
+
+原始模型权重下载地址 :
+
+https://obs-9be7.obs.myhuaweicloud.com/resnet50/resnet50_tensorflow_1.7.pb
+
+
+## om模型
+
+om模型下载地址：
+
+https://obs-9be7.obs.cn-east-2.myhuaweicloud.com/resnet50/tf_resnet50.om
+
+使用ATC模型转换工具进行模型转换时可以参考如下指令，具体操作详情和参数设置可以参考  [ATC工具使用指导](https://support.huaweicloud.com/ti-atc-A200dk_3000/altasatc_16_002.html) 
+
+```
+wget https://obs-9be7.obs.myhuaweicloud.com/models/resnet50_imagenet_dynamic_hw-python/aipp_resnet50.aippconfig
+```
+
+```
+atc --model=./resnet50_tensorflow_1.7.pb  --framework=3 --output=./tf_resnet50 --soc_version=Ascend310  --input_shape="Placeholder:1,-1,-1,3"  --dynamic_image_size="112,112;224,224;448,448" --insert_op_conf=./aipp_resnet50.aippconfig
+```
+
+## 使用msame工具推理
+
+参考 https://gitee.com/ascend/tools/tree/master/msame, 获取msame推理工具及使用方法。
+
+获取到msame可执行文件之后，将待检测om文件放在model文件夹，然后进行性能测试。
+
+## 性能测试
+
+使用msame推理工具，发起推理性能测试。可以参考如下指令： 
+
+```
+ ./msame --model ../../../models/tf_resnet50.om --output output/ --loop 100
+```
+
+性能测试数据为：
+
+```
+[INFO] acl init success
+[INFO] open device 0 success
+[INFO] create context success
+[INFO] create stream success
+[INFO] get run mode success
+[INFO] load model ./tf_resnet50.om success
+[INFO] create model description success
+[INFO] get input dynamic gear count success
+[INFO] create model output success
+output//2022922_17_4_9_955610
+[INFO] model execute success
+Inference time: 8.69ms
+[INFO] get max dynamic batch size success
+[INFO] output data success
+Inference average time: 8.690000 ms
+[INFO] destroy model input success
+[INFO] unload model success, model Id is 1
+[INFO] Execute sample success
+[INFO] end to destroy stream
+[INFO] end to destroy context
+[INFO] end to reset device is 0
+[INFO] end to finalize acl
+```
+
+Batch: 1, shape:  448 *448 *3，带有AIPP，平均推理性能 8.69ms
+
+## 精度测试
+
+待完善
-- 
Gitee