m53297601
/
mindspore
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

deeplabv3 & ssd & lstm new interface

Browse Source 
modified:   model_zoo/official/cv/ssd/README_CN.md
pull/13830/head
unknown 4 years ago
parent f715188478
commit e063925b9c
9 changed files with 109 additions and 36 deletions
Show Stats Download Patch File Download Diff File

6
model_zoo/official/cv/deeplabv3/ascend310_infer/build.sh
Unescape View File

@ -13,11 +13,11 @@
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
if [ ! -d out ]; then
mkdir out
if [ -d out ]; then
rm -rf out
fi
cd out
mkdir out && cd out
if [ -f "Makefile" ]; then
make clean

20
model_zoo/official/cv/deeplabv3/ascend310_infer/src/main.cc
Unescape View File

@ -31,10 +31,9 @@
#include "include/minddata/dataset/include/execute.h"
#include "../inc/utils.h"
using mindspore::GlobalContext;
using mindspore::Context;
using mindspore::Serialization;
using mindspore::Model;
using mindspore::ModelContext;
using mindspore::Status;
using mindspore::ModelType;
using mindspore::GraphCell;
@ -127,15 +126,18 @@ int main(int argc, char **argv) {
std::cout << "Invalid fusion switch path" << std::endl;
return 1;
}
GlobalContext::SetGlobalDeviceTarget(mindspore::kDeviceTypeAscend310);
GlobalContext::SetGlobalDeviceID(FLAGS_device_id);
auto graph = Serialization::LoadModel(FLAGS_mindir_path, ModelType::kMindIR);
auto model_context = std::make_shared<mindspore::Context>();
auto context = std::make_shared<Context>();
auto ascend310 = std::make_shared<mindspore::Ascend310DeviceInfo>();
ascend310->SetDeviceID(FLAGS_device_id);
context->MutableDeviceInfo().push_back(ascend310);
mindspore::Graph graph;
Serialization::Load(FLAGS_mindir_path, ModelType::kMindIR, &graph);
if (!FLAGS_fusion_switch_path.empty()) {
ModelContext::SetFusionSwitchConfigPath(model_context, FLAGS_fusion_switch_path);
ascend310->SetFusionSwitchConfigPath(FLAGS_fusion_switch_path);
}
Model model(GraphCell(graph), model_context);
Status ret = model.Build();
Model model;
Status ret = model.Build(GraphCell(graph), context);
if (ret != kSuccess) {
std::cout << "ERROR: Build failed." << std::endl;
return 1;

2
model_zoo/official/cv/deeplabv3/scripts/run_infer_310.sh
Unescape View File

@ -15,7 +15,7 @@
# ============================================================================
if [[ $# -lt 4 || $# -gt 5 ]]; then
echo "Usage: sh run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DATA_ROOT] [DATA_LIST] [DEVICE_ID]
echo "Usage: bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DATA_ROOT] [DATA_LIST] [DEVICE_ID]
DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
exit 1
fi

2
model_zoo/official/cv/ssd/README.md
Unescape View File

@ -406,7 +406,7 @@ The ckpt_file parameter is required,
### Infer on Ascend310
Before performing inference, the mindir file must bu exported by export script on the 910 environment. We only provide an example of inference using MINDIR model.
Before performing inference, the mindir file must bu exported by `export.py` script. We only provide an example of inference using MINDIR model.
Current batch_Size can only be set to 1. The precision calculation process needs about 70G+ memory space, otherwise the process will be killed for execeeding memory limits.
```shell

47
model_zoo/official/cv/ssd/README_CN.md
Unescape View File

@ -17,6 +17,10 @@
- [评估过程](#评估过程)
- [Ascend处理器环境评估](#ascend处理器环境评估)
- [GPU处理器环境评估](#gpu处理器环境评估)
- [推理过程](#推理过程)
- [导出MindIR](#导出mindir)
- [在Ascend310执行推理](#在ascend310执行推理)
- [结果](#结果)
- [模型描述](#模型描述)
- [性能](#性能)
- [评估性能](#评估性能)
@ -312,6 +316,49 @@ Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.686
mAP: 0.2244936111705981
```
## 推理过程
### [导出MindIR](#contents)
```shell
python export.py --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT]
```
参数ckpt_file为必填项，
`EXPORT_FORMAT` 必须在 ["AIR", "MINDIR"]中选择。
### 在Ascend310执行推理
在执行推理前，mindir文件必须通过`export.py`脚本导出。以下展示了使用minir模型执行推理的示例。
目前仅支持batch_Size为1的推理。精度计算过程需要70G+的内存，否则进程将会因为超出内存被系统终止。
```shell
# Ascend310 inference
bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [DEVICE_ID]
```
- `DVPP` 为必填项，需要在["DVPP", "CPU"]选择，大小写均可。需要注意的是ssd_vgg16执行推理的图片尺寸为[300, 300]，由于DVPP硬件限制宽为16整除，高为2整除，因此，这个网络需要通过CPU算子对图像进行前处理。
- `DEVICE_ID` 可选，默认值为0。
### 结果
推理结果保存在脚本执行的当前路径，你可以在acc.log中看到以下精度计算结果。
```bash
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.339
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.521
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.370
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.168
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.386
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.461
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.310
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.481
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.515
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.293
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.659
mAP: 0.33880018942412393
```
# 模型描述
## 性能

6
model_zoo/official/cv/ssd/ascend310_infer/build.sh
Unescape View File

@ -13,11 +13,11 @@
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
if [ ! -d out ]; then
mkdir out
if [ -d out ]; then
rm -rf out
fi
cd out
mkdir out && cd out
if [ -f "Makefile" ]; then
make clean

19
model_zoo/official/cv/ssd/ascend310_infer/src/main.cc
Unescape View File

@ -32,10 +32,9 @@
#include "include/minddata/dataset/include/vision.h"
#include "inc/utils.h"
using mindspore::GlobalContext;
using mindspore::Context;
using mindspore::Serialization;
using mindspore::Model;
using mindspore::ModelContext;
using mindspore::Status;
using mindspore::ModelType;
using mindspore::GraphCell;
@ -64,21 +63,23 @@ int main(int argc, char **argv) {
return 1;
}
GlobalContext::SetGlobalDeviceTarget(mindspore::kDeviceTypeAscend310);
GlobalContext::SetGlobalDeviceID(FLAGS_device_id);
auto graph = Serialization::LoadModel(FLAGS_mindir_path, ModelType::kMindIR);
auto model_context = std::make_shared<mindspore::ModelContext>();
auto context = std::make_shared<Context>();
auto ascend310 = std::make_shared<mindspore::Ascend310DeviceInfo>();
ascend310->SetDeviceID(FLAGS_device_id);
context->MutableDeviceInfo().push_back(ascend310);
mindspore::Graph graph;
Serialization::Load(FLAGS_mindir_path, ModelType::kMindIR, &graph);
if (FLAGS_cpu_dvpp == "DVPP") {
if (RealPath(FLAGS_aipp_path).empty()) {
std::cout << "Invalid aipp path" << std::endl;
return 1;
} else {
ModelContext::SetInsertOpConfigPath(model_context, FLAGS_aipp_path);
ascend310->SetInsertOpConfigPath(FLAGS_aipp_path);
}
}
Model model(GraphCell(graph), model_context);
Status ret = model.Build();
Model model;
Status ret = model.Build(GraphCell(graph), context);
if (ret != kSuccess) {
std::cout << "ERROR: Build failed." << std::endl;
return 1;

4
model_zoo/official/cv/ssd/scripts/run_infer_310.sh
Unescape View File

@ -15,7 +15,7 @@
# ============================================================================
if [[ $# -lt 3 || $# -gt 4 ]]; then
echo "Usage: sh run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [DEVICE_ID]
echo "Usage: bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [DEVICE_ID]
DVPP is mandatory, and must choose from [DVPP|CPU], it's case-insensitive
DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
exit 1
@ -59,7 +59,7 @@ fi
function compile_app()
{
cd ../ascend310_infer
sh build.sh &> build.log
bash build.sh &> build.log
}
function infer()

33
model_zoo/official/nlp/lstm/export.py
Unescape View File

@ -1,4 +1,4 @@
# Copyright 2020 Huawei Technologies Co., Ltd
# Copyright 2020-2021 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
@ -21,9 +21,9 @@ import os
import numpy as np
from mindspore import Tensor
from mindspore import Tensor, context
from mindspore import export, load_checkpoint, load_param_into_net
from src.config import lstm_cfg as cfg
from src.config import lstm_cfg, lstm_cfg_ascend
from src.lstm import SentimentNet
if __name__ == '__main__':
@ -31,9 +31,32 @@ if __name__ == '__main__':
parser.add_argument('--preprocess_path', type=str, default='./preprocess',
help='path where the pre-process data is stored.')
parser.add_argument('--ckpt_file', type=str, required=True, help='lstm ckpt file.')
parser.add_argument("--device_id", type=int, default=0, help="Device id")
parser.add_argument("--file_name", type=str, default="lstm", help="output file name.")
parser.add_argument('--file_format', type=str, choices=["AIR", "MINDIR"], default='AIR', help='file format')
parser.add_argument('--device_target', type=str, default="Ascend", choices=['GPU', 'CPU', 'Ascend'],
help='the target device to run, support "GPU", "CPU". Default: "Ascend".')
args = parser.parse_args()
context.set_context(
mode=context.GRAPH_MODE,
save_graphs=False,
device_target=args.device_target,
device_id=args.device_id)
if args.device_target == 'Ascend':
cfg = lstm_cfg_ascend
else:
cfg = lstm_cfg
embedding_table = np.loadtxt(os.path.join(args.preprocess_path, "weight.txt")).astype(np.float32)
if args.device_target == 'Ascend':
pad_num = int(np.ceil(cfg.embed_size / 16) * 16 - cfg.embed_size)
if pad_num > 0:
embedding_table = np.pad(embedding_table, [(0, 0), (0, pad_num)], 'constant')
cfg.embed_size = int(np.ceil(cfg.embed_size / 16) * 16)
network = SentimentNet(vocab_size=embedding_table.shape[0],
embed_size=cfg.embed_size,
num_hiddens=cfg.num_hiddens,
@ -46,5 +69,5 @@ if __name__ == '__main__':
param_dict = load_checkpoint(args.ckpt_file)
load_param_into_net(network, param_dict)
input_arr = Tensor(np.random.uniform(0.0, 1e5, size=[64, 500]).astype(np.int32))
export(network, input_arr, file_name="lstm", file_format="MINDIR")
input_arr = Tensor(np.random.uniform(0.0, 1e5, size=[cfg.batch_size, 500]).astype(np.int32))
export(network, input_arr, file_name=args.file_name, file_format=args.file_format)

Write Preview
Loading…
Cancel
Save