Skip to content

GitLab

"...data/git@developer.sourcefind.cn:wangsen/paddle_dbnet.git" did not exist on "e885b57ea5a6cec66982935db9f34ad8ad329820"
Commit 237febb1 authored by LDOUBLEV's avatar LDOUBLEV
Browse files

fix conflict

parents 4cb2179b 0b37c118
Hide whitespace changes
Inline Side-by-side
Showing with 729 additions and 200 deletions
deploy/paddle2onnx/readme.md 0 → 100644
View file @ 237febb1
# paddle2onnx 模型转化与预测
本章节介绍 PaddleOCR 模型如何转化为 ONNX 模型，并基于 ONNX 引擎预测。
## 1. 环境准备
需要准备 Paddle2ONNX 模型转化环境，和 ONNX 模型预测环境
### Paddle2ONNX
Paddle2ONNX 支持将 PaddlePaddle 模型格式转化到 ONNX 模型格式，算子目前稳定支持导出 ONNX Opset 9~11，部分Paddle算子支持更低的ONNX Opset转换。
更多细节可参考 [Paddle2ONNX](https://github.com/PaddlePaddle/Paddle2ONNX/blob/develop/README_zh.md)
- 安装 Paddle2ONNX
```
python3.7 -m pip install paddle2onnx
```
- 安装 ONNX
```
# 建议安装 1.4.0 版本，可根据环境更换版本号
python3.7 -m pip install onnxruntime==1.4.0
```
## 2. 模型转换
- Paddle 模型下载
有两种方式获取Paddle静态图模型：在 [model_list](../../doc/doc_ch/models_list.md) 中下载PaddleOCR提供的预测模型；
参考[模型导出说明](../../doc/doc_ch/inference.md#训练模型转inference模型)把训练好的权重转为 inference_model。
以 ppocr 检测模型为例：
```
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar
cd ./inference && tar xf ch_ppocr_mobile_v2.0_det_infer.tar && cd ..
```
- 模型转换
使用 Paddle2ONNX 将Paddle静态图模型转换为ONNX模型格式：
```
paddle2onnx --model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ \
--model_filename=inference.pdmodel \
--params_filename=inference.pdiparams \
--save_file=./inference/det_mobile_onnx/model.onnx \
--opset_version=10 \
--enable_onnx_checker=True
```
执行完毕后，ONNX 模型会被保存在 `./inference/det_mobile_onnx/` 路径下
## 3. onnx 预测
以检测模型为例，使用 ONNX 预测可执行如下命令：
```
python3.7 ../../tools/infer/predict_det.py --use_gpu=False --use_onnx=True \
--det_model_dir=./inference/det_mobile_onnx/model.onnx \
--image_dir=../../doc/imgs/1.jpg
```
执行命令后在终端会打印出预测的检测框坐标，并在 `./inference_results/` 下保存可视化结果。
```
root INFO: 1.jpg [[[291, 295], [334, 292], [348, 844], [305, 847]], [[344, 296], [379, 294], [387, 669], [353, 671]]]
The predict time of ../../doc/imgs/1.jpg: 0.06162881851196289
The visualized image saved in ./inference_results/det_res_1.jpg
```
* 注意：ONNX暂时不支持变长预测，因为需要将输入resize到固定输入，预测结果可能与直接使用Paddle预测有细微不同。
doc/doc_ch/algorithm_overview.md
View file @ 237febb1
......@@ -68,4 +68,5 @@ PaddleOCR基于动态图开源的文本识别算法列表：
|NRTR|NRTR_MTB| 84.3% | rec_mtb_nrtr | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mtb_nrtr_train.tar) |
|SAR|Resnet31| 87.2% | rec_r31_sar | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.1/rec/rec_r31_sar_train.tar) |
|SEED| Aster_Resnet | 85.2% | rec_resnet_stn_bilstm_att | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.1/rec/rec_resnet_stn_bilstm_att.tar)|
PaddleOCR文本识别算法的训练和使用请参考文档教程中[模型训练/评估中的文本识别部分](./recognition.md)。
test_tipc/configs/amp_ppocr_det_mobile_params.txt 0 → 100644
View file @ 237febb1
===========================train_params===========================
model_name:ocr_det
python:python3.7
gpu_list:0|0,1
Global.use_gpu:True|True
Global.auto_cast:amp
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=300
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=2|whole_train_whole_infer=4
Global.pretrained_model:null
train_model_name:latest
train_infer_img_dir:./train_data/icdar2015/text_localization/ch4_test_images/
null:null
##
trainer:norm_train|pact_train|fpgm_train
norm_train:tools/train.py -c test_tipc/configs/det_mv3_db.yml -o Global.pretrained_model=./pretrain_models/MobileNetV3_large_x0_5_pretrained
pact_train:deploy/slim/quantization/quant.py -c test_tipc/configs/det_mv3_db.yml -o
fpgm_train:deploy/slim/prune/sensitivity_anal.py -c test_tipc/configs/det_mv3_db.yml -o Global.pretrained_model=./pretrain_models/det_mv3_db_v2.0_train/best_accuracy
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.pretrained_model:
norm_export:tools/export_model.py -c test_tipc/configs/det_mv3_db.yml -o
quant_export:deploy/slim/quantization/export_model.py -c test_tipc/configs/det_mv3_db.yml -o
fpgm_export:deploy/slim/prune/export_prune_model.py -c test_tipc/configs/det_mv3_db.yml -o
distill_export:null
export1:null
export2:null
inference_dir:null
train_model:./inference/ch_ppocr_mobile_v2.0_det_train/best_accuracy
infer_export:tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
infer_quant:False
inference:tools/infer/predict_det.py
--use_gpu:True|False
--enable_mkldnn:True|False
--cpu_threads:1|6
--rec_batch_num:1
--use_tensorrt:False|True
--precision:fp32|fp16|int8
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
null:null
--benchmark:True
null:null
===========================cpp_infer_params===========================
use_opencv:True
infer_model:./inference/ch_ppocr_mobile_v2.0_det_infer/
infer_quant:False
inference:./deploy/cpp_infer/build/ppocr det
--use_gpu:True|False
--enable_mkldnn:True|False
--cpu_threads:1|6
--rec_batch_num:1
--use_tensorrt:False|True
--precision:fp32|fp16
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
null:null
--benchmark:True
===========================serving_params===========================
model_name:ocr_det
python:python3.7
trans_model:-m paddle_serving_client.convert
--dirname:./inference/ch_ppocr_mobile_v2.0_det_infer/
--model_filename:inference.pdmodel
--params_filename:inference.pdiparams
--serving_server:./deploy/pdserving/ppocr_det_mobile_2.0_serving/
--serving_client:./deploy/pdserving/ppocr_det_mobile_2.0_client/
serving_dir:./deploy/pdserving
web_service:web_service_det.py --config=config.yml --opt op.det.concurrency=1
op.det.local_service_conf.devices:null|0
op.det.local_service_conf.use_mkldnn:True|False
op.det.local_service_conf.thread_num:1|6
op.det.local_service_conf.use_trt:False|True
op.det.local_service_conf.precision:fp32|fp16|int8
pipline:pipeline_http_client.py --image_dir=../../doc/imgs
===========================kl_quant_params===========================
infer_model:./inference/ch_ppocr_mobile_v2.0_det_infer/
infer_export:tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
infer_quant:True
inference:tools/infer/predict_det.py
--use_gpu:True|False
--enable_mkldnn:True|False
--cpu_threads:1|6
--rec_batch_num:1
--use_tensorrt:False|True
--precision:int8
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
null:null
--benchmark:True
null:null
null:null
===========================lite_params===========================
inference:./ocr_db_crnn det
infer_model:./models/ch_ppocr_mobile_v2.0_det_opt.nb|./models/ch_ppocr_mobile_v2.0_det_slim_opt.nb
--cpu_threads:1|4
--batch_size:1
--power_mode:LITE_POWER_HIGH|LITE_POWER_LOW
--image_dir:./test_data/icdar2015_lite/text_localization/ch4_test_images/|./test_data/icdar2015_lite/text_localization/ch4_test_images/img_233.jpg
--config_dir:./config.txt
--rec_dict_dir:./ppocr_keys_v1.txt
--benchmark:True
test_tipc/configs/fleet_ppocr_det_mobile_params.txt 0 → 100644
View file @ 237febb1
===========================train_params===========================
model_name:ocr_det
python:python3.7
gpu_list:xx.xx.xx.xx,xx.xx.xx.xx;0,1
Global.use_gpu:True|True
Global.auto_cast:null|amp
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=300
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=2|whole_train_whole_infer=4
Global.pretrained_model:null
train_model_name:latest
train_infer_img_dir:./train_data/icdar2015/text_localization/ch4_test_images/
null:null
##
trainer:norm_train|pact_train|fpgm_train
norm_train:tools/train.py -c test_tipc/configs/det_mv3_db.yml -o Global.pretrained_model=./pretrain_models/MobileNetV3_large_x0_5_pretrained
pact_train:deploy/slim/quantization/quant.py -c test_tipc/configs/det_mv3_db.yml -o
fpgm_train:deploy/slim/prune/sensitivity_anal.py -c test_tipc/configs/det_mv3_db.yml -o Global.pretrained_model=./pretrain_models/det_mv3_db_v2.0_train/best_accuracy
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.pretrained_model:
norm_export:tools/export_model.py -c test_tipc/configs/det_mv3_db.yml -o
quant_export:deploy/slim/quantization/export_model.py -c test_tipc/configs/det_mv3_db.yml -o
fpgm_export:deploy/slim/prune/export_prune_model.py -c test_tipc/configs/det_mv3_db.yml -o
distill_export:null
export1:null
export2:null
inference_dir:null
train_model:./inference/ch_ppocr_mobile_v2.0_det_train/best_accuracy
infer_export:tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
infer_quant:False
inference:tools/infer/predict_det.py
--use_gpu:True|False
--enable_mkldnn:True|False
--cpu_threads:1|6
--rec_batch_num:1
--use_tensorrt:False|True
--precision:fp32|fp16|int8
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
null:null
--benchmark:True
null:null
===========================cpp_infer_params===========================
use_opencv:True
infer_model:./inference/ch_ppocr_mobile_v2.0_det_infer/
infer_quant:False
inference:./deploy/cpp_infer/build/ppocr det
--use_gpu:True|False
--enable_mkldnn:True|False
--cpu_threads:1|6
--rec_batch_num:1
--use_tensorrt:False|True
--precision:fp32|fp16
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
null:null
--benchmark:True
===========================serving_params===========================
model_name:ocr_det
python:python3.7
trans_model:-m paddle_serving_client.convert
--dirname:./inference/ch_ppocr_mobile_v2.0_det_infer/
--model_filename:inference.pdmodel
--params_filename:inference.pdiparams
--serving_server:./deploy/pdserving/ppocr_det_mobile_2.0_serving/
--serving_client:./deploy/pdserving/ppocr_det_mobile_2.0_client/
serving_dir:./deploy/pdserving
web_service:web_service_det.py --config=config.yml --opt op.det.concurrency=1
op.det.local_service_conf.devices:null|0
op.det.local_service_conf.use_mkldnn:True|False
op.det.local_service_conf.thread_num:1|6
op.det.local_service_conf.use_trt:False|True
op.det.local_service_conf.precision:fp32|fp16|int8
pipline:pipeline_http_client.py --image_dir=../../doc/imgs
===========================kl_quant_params===========================
infer_model:./inference/ch_ppocr_mobile_v2.0_det_infer/
infer_export:tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
infer_quant:True
inference:tools/infer/predict_det.py
--use_gpu:True|False
--enable_mkldnn:True|False
--cpu_threads:1|6
--rec_batch_num:1
--use_tensorrt:False|True
--precision:int8
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
null:null
--benchmark:True
null:null
null:null
===========================lite_params===========================
inference:./ocr_db_crnn det
infer_model:./models/ch_ppocr_mobile_v2.0_det_opt.nb|./models/ch_ppocr_mobile_v2.0_det_slim_opt.nb
--cpu_threads:1|4
--batch_size:1
--power_mode:LITE_POWER_HIGH|LITE_POWER_LOW
--image_dir:./test_data/icdar2015_lite/text_localization/ch4_test_images/|./test_data/icdar2015_lite/text_localization/ch4_test_images/img_233.jpg
--config_dir:./config.txt
--rec_dict_dir:./ppocr_keys_v1.txt
--benchmark:True
test_tipc/configs/ppocr_det_mobile_params.txt
View file @ 237febb1
===========================train_params===========================
model_name:ocr_det
python:python3.7
gpu_list:0|0,1|10.21.226.181,10.21.226.133;0,1
Global.use_gpu:True|True|True
Global.auto_cast:fp32|amp
gpu_list:0|0,1
Global.use_gpu:True|True
Global.auto_cast:null
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=300
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=2|whole_train_whole_infer=4
......@@ -108,3 +108,15 @@ infer_model:./models/ch_ppocr_mobile_v2.0_det_opt.nb|./models/ch_ppocr_mobile_v2
--config_dir:./config.txt
--rec_dict_dir:./ppocr_keys_v1.txt
--benchmark:True
===========================paddle2onnx_params===========================
2onnx: paddle2onnx
--model_dir:./inference/ch_ppocr_mobile_v2.0_det_infer/
--model_filename:inference.pdmodel
--params_filename:inference.pdiparams
--save_file:./inference/det_mobile_onnx/model.onnx
--opset_version:10
--enable_onnx_checker:True
inference:tools/infer/predict_det.py
--use_gpu:False
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
\ No newline at end of file
test_tipc/docs/test_inference_cpp.md
View file @ 237febb1
......@@ -14,6 +14,8 @@ C++预测功能测试的主程序为`test_inference_cpp.sh`，可以测试基于
| 量化模型 | CPU | 1/6 | - | int8 | 支持 |
## 2. 测试流程
运行环境配置请参考[文档](./install.md)的内容配置TIPC的运行环境。
### 2.1 功能测试
先运行`prepare.sh`准备数据和模型，然后运行`test_inference_cpp.sh`进行测试，最终在```test_tipc/output```目录下生成`cpp_infer_*.log`后缀的日志文件。
......@@ -26,6 +28,32 @@ bash test_tipc/test_inference_cpp.sh ./test_tipc/configs/ppocr_det_mobile_params
bash test_tipc/test_inference_cpp.sh ./test_tipc/configs/ppocr_det_mobile_params.txt '1'
```
运行预测指令后，在`test_tipc/output`文件夹下自动会保存运行日志，包括以下文件：
```shell
test_tipc/output/
|- results_cpp.log # 运行指令状态的日志
|- cpp_infer_cpu_usemkldnn_False_threads_1_precision_fp32_batchsize_1.log # CPU上不开启Mkldnn，线程数设置为1，测试batch_size=1条件下的预测运行日志
|- cpp_infer_cpu_usemkldnn_False_threads_6_precision_fp32_batchsize_1.log # CPU上不开启Mkldnn，线程数设置为6，测试batch_size=1条件下的预测运行日志
|- cpp_infer_gpu_usetrt_False_precision_fp32_batchsize_1.log # GPU上不开启TensorRT，测试batch_size=1的fp32精度预测日志
|- cpp_infer_gpu_usetrt_True_precision_fp16_batchsize_1.log # GPU上开启TensorRT，测试batch_size=1的fp16精度预测日志
......
```
其中results_cpp.log中包含了每条指令的运行状态，如果运行成功会输出：
```
Run successfully with command - ./deploy/cpp_infer/build/ppocr det --use_gpu=False --enable_mkldnn=False --cpu_threads=6 --det_model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ --rec_batch_num=1 --image_dir=./inference/ch_det_data_50/all-sum-510/ --benchmar k=True > ./test_tipc/output/cpp_infer_cpu_usemkldnn_False_threads_6_precision_fp32_batchsize_1.log 2>&1 !
Run successfully with command - ./deploy/cpp_infer/build/ppocr det --use_gpu=True --use_tensorrt=False --precision=fp32 --det_model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ --rec_batch_num=1 --image_dir=./inference/ch_det_data_50/all-sum-510/ --benchmark =True > ./test_tipc/output/cpp_infer_gpu_usetrt_False_precision_fp32_batchsize_1.log 2>&1 !
......
```
如果运行失败，会输出：
```
Run failed with command - ./deploy/cpp_infer/build/ppocr det --use_gpu=True --use_tensorrt=True --precision=fp32 --det_model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ --rec_batch_num=1 --image_dir=./inference/ch_det_data_50/all-sum-510/ --benchmark=True > ./test_tipc/output/cpp_infer_gpu_usetrt_True_precision_fp32_batchsize_1.log 2>&1 !
Run failed with command - ./deploy/cpp_infer/build/ppocr det --use_gpu=True --use_tensorrt=True --precision=fp16 --det_model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ --rec_batch_num=1 --image_dir=./inference/ch_det_data_50/all-sum-510/ --benchmark=True > ./test_tipc/output/cpp_infer_gpu_usetrt_True_precision_fp16_batchsize_1.log 2>&1 !
......
```
可以很方便的根据results_cpp.log中的内容判定哪一个指令运行错误。
### 2.2 精度测试
......
test_tipc/docs/test_lite.md
View file @ 237febb1
......@@ -20,6 +20,7 @@ Lite预测功能测试的主程序为`test_lite.sh`，可以测试基于Lite预
## 2. 测试流程
运行环境配置请参考[文档](./install.md)的内容配置TIPC的运行环境。
### 2.1 功能测试
......
test_tipc/docs/test_paddle2onnx.md 0 → 100644
View file @ 237febb1
# Paddle2onnx预测功能测试
PaddleServing预测功能测试的主程序为`test_paddle2onnx.sh`，可以测试Paddle2ONNX的模型转化功能，并验证正确性。
## 1. 测试结论汇总
基于训练是否使用量化，进行本测试的模型可以分为`正常模型`和`量化模型`，这两类模型对应的Paddle2ONNX预测功能汇总如下：
| 模型类型 |device |
| ---- | ---- |
| 正常模型 | GPU |
| 正常模型 | CPU |
| 量化模型 | GPU |
| 量化模型 | CPU |
## 2. 测试流程
### 2.1 功能测试
先运行`prepare.sh`准备数据和模型，然后运行`test_paddle2onnx.sh`进行测试，最终在```test_tipc/output```目录下生成`paddle2onnx_infer_*.log`后缀的日志文件。
```shell
bash test_tipc/prepare.sh ./test_tipc/configs/ppocr_det_mobile_params.txt "paddle2onnx_infer"
# 用法:
bash test_tipc/test_paddle2onnx.sh ./test_tipc/configs/ppocr_det_mobile_params.txt
```
#### 运行结果
各测试的运行情况会打印在 `test_tipc/output/results_paddle2onnx.log` 中：
运行成功时会输出：
```
Run successfully with command - paddle2onnx --model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ --model_filename=inference.pdmodel --params_filename=inference.pdiparams --save_file=./inference/det_mobile_onnx/model.onnx --opset_version=10 --enable_onnx_checker=True!
Run successfully with command - python test_tipc/onnx_inference/predict_det.py --use_gpu=False --image_dir=./inference/ch_det_data_50/all-sum-510/ --det_model_dir=./inference/det_mobile_onnx/model.onnx 2>&1 !
```
运行失败时会输出：
```
Run failed with command - paddle2onnx --model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ --model_filename=inference.pdmodel --params_filename=inference.pdiparams --save_file=./inference/det_mobile_onnx/model.onnx --opset_version=10 --enable_onnx_checker=True!
...
```
## 3. 更多教程
本文档为功能测试用，更详细的Paddle2onnx预测使用教程请参考：[Paddle2ONNX](https://github.com/PaddlePaddle/Paddle2ONNX)
test_tipc/docs/test_serving.md
View file @ 237febb1
......@@ -4,7 +4,7 @@ PaddleServing预测功能测试的主程序为`test_serving.sh`，可以测试
## 1. 测试结论汇总
基于训练是否使用量化，进行本测试的模型可以分为`正常模型`和`量化模型`，这两类模型对应的C++预测功能汇总如下：
基于训练是否使用量化，进行本测试的模型可以分为`正常模型`和`量化模型`，这两类模型对应的Serving预测功能汇总如下：
| 模型类型 |device | batchsize | tensorrt | mkldnn | cpu多线程 |
| ---- | ---- | ---- | :----: | :----: | :----: |
......@@ -14,6 +14,8 @@ PaddleServing预测功能测试的主程序为`test_serving.sh`，可以测试
| 量化模型 | CPU | 1/6 | - | int8 | 支持 |
## 2. 测试流程
运行环境配置请参考[文档](./install.md)的内容配置TIPC的运行环境。
### 2.1 功能测试
先运行`prepare.sh`准备数据和模型，然后运行`test_serving.sh`进行测试，最终在```test_tipc/output```目录下生成`serving_infer_*.log`后缀的日志文件。
......
test_tipc/docs/test_train_inference_python.md
View file @ 237febb1
......@@ -32,7 +32,7 @@ Linux端基础训练预测功能测试的主程序为`test_train_inference_pytho
## 2. 测试流程
运行环境配置请参考[文档](./install.md)的内容配置tipc的运行环境。
运行环境配置请参考[文档](./install.md)的内容配置TIPC的运行环境。
### 2.1 安装依赖
- 安装PaddlePaddle >= 2.0
......
test_tipc/docs/win_test_train_inference_python.md
View file @ 237febb1
......@@ -23,7 +23,7 @@ Windows端基础训练预测功能测试的主程序为`test_train_inference_pyt
## 2. 测试流程
运行环境配置请参考[文档](./install.md)的内容配置tipc的运行环境。
运行环境配置请参考[文档](./install.md)的内容配置TIPC的运行环境。
另外，由于Windows上和linux的路径管理方式不同，可以在win上安装gitbash终端，在gitbash中执行指令的方式和在linux端执行指令方式相同，更方便tipc测试。gitbash[下载链接](https://git-scm.com/download/win)。
......
test_tipc/prepare.sh
View file @ 237febb1
......@@ -202,3 +202,20 @@ if [ ${MODE} = "lite_infer" ];then
tar -cf test_lite.tar ./test_lite && cp test_lite.tar ${current_dir} && cd ${current_dir}
fi
if [ ${MODE} = "paddle2onnx_infer" ];then
# prepare serving env
python_name=$(func_parser_value "${lines[2]}")
${python_name} -m pip install install paddle2onnx
${python_name} -m pip install onnxruntime==1.4.0
# wget model
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar
# wget data
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/ch_det_data_50.tar
wget -nc -P ./inference/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/rec_inference.tar
cd ./inference && tar xf ch_ppocr_mobile_v2.0_det_infer.tar && tar xf ch_ppocr_mobile_v2.0_rec_infer.tar && tar xf ch_ppocr_server_v2.0_rec_infer.tar && tar xf ch_ppocr_server_v2.0_det_infer.tar && tar xf ch_det_data_50.tar && tar xf rec_inference.tar && cd ../
fi
test_tipc/test_paddle2onnx.sh 0 → 100644
View file @ 237febb1
#!/bin/bash
source test_tipc/common_func.sh
FILENAME=$1
dataline=$(cat ${FILENAME})
lines=(${dataline})
# common params
model_name=$(func_parser_value "${lines[1]}")
python=$(func_parser_value "${lines[2]}")
# parser params
dataline=$(awk 'NR==111, NR==123{print}' $FILENAME)
IFS=$'\n'
lines=(${dataline})
# parser paddle2onnx
padlle2onnx_cmd=$(func_parser_value "${lines[1]}")
infer_model_dir_key=$(func_parser_key "${lines[2]}")
infer_model_dir_value=$(func_parser_value "${lines[2]}")
model_filename_key=$(func_parser_key "${lines[3]}")
model_filename_value=$(func_parser_value "${lines[3]}")
params_filename_key=$(func_parser_key "${lines[4]}")
params_filename_value=$(func_parser_value "${lines[4]}")
save_file_key=$(func_parser_key "${lines[5]}")
save_file_value=$(func_parser_value "${lines[5]}")
opset_version_key=$(func_parser_key "${lines[6]}")
opset_version_value=$(func_parser_value "${lines[6]}")
enable_onnx_checker_key=$(func_parser_key "${lines[7]}")
enable_onnx_checker_value=$(func_parser_value "${lines[7]}")
# parser onnx inference
inference_py=$(func_parser_value "${lines[8]}")
use_gpu_key=$(func_parser_key "${lines[9]}")
use_gpu_value=$(func_parser_value "${lines[9]}")
det_model_key=$(func_parser_key "${lines[10]}")
image_dir_key=$(func_parser_key "${lines[11]}")
image_dir_value=$(func_parser_value "${lines[11]}")
LOG_PATH="./test_tipc/output"
mkdir -p ./test_tipc/output
status_log="${LOG_PATH}/results_paddle2onnx.log"
function func_paddle2onnx(){
IFS='|'
_script=$1
# paddle2onnx
_save_log_path="${LOG_PATH}/paddle2onnx_infer_cpu.log"
set_dirname=$(func_set_params "${infer_model_dir_key}" "${infer_model_dir_value}")
set_model_filename=$(func_set_params "${model_filename_key}" "${model_filename_value}")
set_params_filename=$(func_set_params "${params_filename_key}" "${params_filename_value}")
set_save_model=$(func_set_params "${save_file_key}" "${save_file_value}")
set_opset_version=$(func_set_params "${opset_version_key}" "${opset_version_value}")
set_enable_onnx_checker=$(func_set_params "${enable_onnx_checker_key}" "${enable_onnx_checker_value}")
trans_model_cmd="${padlle2onnx_cmd} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_save_model} ${set_opset_version} ${set_enable_onnx_checker}"
eval $trans_model_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${trans_model_cmd}" "${status_log}"
# python inference
set_gpu=$(func_set_params "${use_gpu_key}" "${use_gpu_value}")
set_model_dir=$(func_set_params "${det_model_key}" "${save_file_value}")
set_img_dir=$(func_set_params "${image_dir_key}" "${image_dir_value}")
infer_model_cmd="${python} ${inference_py} ${set_gpu} ${set_img_dir} ${set_model_dir} --use_onnx=True > ${_save_log_path} 2>&1 "
eval $infer_model_cmd
status_check $last_status "${infer_model_cmd}" "${status_log}"
}
echo "################### run test ###################"
export Count=0
IFS="|"
func_paddle2onnx
\ No newline at end of file
test_tipc/test_train_inference_python.sh
View file @ 237febb1
......@@ -316,7 +316,7 @@ else
elif [ ${#ips} -le 26 ];then # train with multi-gpu
cmd="${python} -m paddle.distributed.launch --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_epoch} ${set_pretrain} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
else # train with multi-machine
cmd="${python} -m paddle.distributed.launch --ips=${ips} --gpus=${gpu} ${set_use_gpu} ${run_train} ${set_save_model} ${set_pretrain} ${set_epoch} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
cmd="${python} -m paddle.distributed.launch --ips=${ips} --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_pretrain} ${set_epoch} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
fi
# run train
eval "unset CUDA_VISIBLE_DEVICES"
......
tools/infer/predict_cls.py
View file @ 237febb1
......@@ -47,6 +47,7 @@ class TextClassifier(object):
self.postprocess_op = build_post_process(postprocess_params)
self.predictor, self.input_tensor, self.output_tensors, _ = \
utility.create_predictor(args, 'cls', logger)
self.use_onnx = args.use_onnx
def resize_norm_img(self, img):
imgC, imgH, imgW = self.cls_image_shape
......@@ -100,10 +101,16 @@ class TextClassifier(object):
norm_img_batch = np.concatenate(norm_img_batch)
norm_img_batch = norm_img_batch.copy()
self.input_tensor.copy_from_cpu(norm_img_batch)
self.predictor.run()
prob_out = self.output_tensors[0].copy_to_cpu()
self.predictor.try_shrink_memory()
if self.use_onnx:
input_dict = {}
input_dict[self.input_tensor.name] = norm_img_batch
outputs = self.predictor.run(self.output_tensors, input_dict)
prob_out = outputs[0]
else:
self.input_tensor.copy_from_cpu(norm_img_batch)
self.predictor.run()
prob_out = self.output_tensors[0].copy_to_cpu()
self.predictor.try_shrink_memory()
cls_result = self.postprocess_op(prob_out)
elapse += time.time() - starttime
for rno in range(len(cls_result)):
......
tools/infer/predict_det.py
View file @ 237febb1
......@@ -38,6 +38,7 @@ class TextDetector(object):
def __init__(self, args):
self.args = args
self.det_algorithm = args.det_algorithm
self.use_onnx = args.use_onnx
pre_process_list = [{
'DetResizeForTest': {
'limit_side_len': args.det_limit_side_len,
......@@ -100,7 +101,12 @@ class TextDetector(object):
else:
logger.info("unknown det_algorithm:{}".format(self.det_algorithm))
sys.exit(0)
if self.use_onnx:
pre_process_list[0] = {
'DetResizeForTest': {
'image_shape': [640, 640]
}
}
self.preprocess_op = create_operators(pre_process_list)
self.postprocess_op = build_post_process(postprocess_params)
self.predictor, self.input_tensor, self.output_tensors, self.config = utility.create_predictor(
......@@ -198,15 +204,19 @@ class TextDetector(object):
if self.args.benchmark:
self.autolog.times.stamp()
self.input_tensor.copy_from_cpu(img)
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.args.benchmark:
self.autolog.times.stamp()
if self.use_onnx:
input_dict = {}
input_dict[self.input_tensor.name] = img
outputs = self.predictor.run(self.output_tensors, input_dict)
else:
self.input_tensor.copy_from_cpu(img)
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.args.benchmark:
self.autolog.times.stamp()
preds = {}
if self.det_algorithm == "EAST":
......
tools/infer/predict_rec.py
View file @ 237febb1
......@@ -73,6 +73,7 @@ class TextRecognizer(object):
self.predictor, self.input_tensor, self.output_tensors, self.config = \
utility.create_predictor(args, 'rec', logger)
self.benchmark = args.benchmark
self.use_onnx = args.use_onnx
if args.benchmark:
import auto_log
pid = os.getpid()
......@@ -107,6 +108,8 @@ class TextRecognizer(object):
assert imgC == img.shape[2]
imgW = int((32 * max_wh_ratio))
if self.use_onnx:
imgW = 100
h, w = img.shape[:2]
ratio = w / float(h)
if math.ceil(imgH * ratio) > imgW:
......@@ -296,51 +299,72 @@ class TextRecognizer(object):
gsrm_slf_attn_bias1_list,
gsrm_slf_attn_bias2_list,
]
input_names = self.predictor.get_input_names()
for i in range(len(input_names)):
input_tensor = self.predictor.get_input_handle(input_names[
i])
input_tensor.copy_from_cpu(inputs[i])
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.benchmark:
self.autolog.times.stamp()
preds = {"predict": outputs[2]}
if self.use_onnx:
input_dict = {}
input_dict[self.input_tensor.name] = norm_img_batch
outputs = self.predictor.run(self.output_tensors,
input_dict)
preds = {"predict": outputs[2]}
else:
input_names = self.predictor.get_input_names()
for i in range(len(input_names)):
input_tensor = self.predictor.get_input_handle(
input_names[i])
input_tensor.copy_from_cpu(inputs[i])
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.benchmark:
self.autolog.times.stamp()
preds = {"predict": outputs[2]}
elif self.rec_algorithm == "SAR":
valid_ratios = np.concatenate(valid_ratios)
inputs = [
norm_img_batch,
valid_ratios,
]
input_names = self.predictor.get_input_names()
for i in range(len(input_names)):
input_tensor = self.predictor.get_input_handle(input_names[
i])
input_tensor.copy_from_cpu(inputs[i])
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.benchmark:
self.autolog.times.stamp()
preds = outputs[0]
else:
self.input_tensor.copy_from_cpu(norm_img_batch)
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.benchmark:
self.autolog.times.stamp()
if len(outputs) != 1:
preds = outputs
if self.use_onnx:
input_dict = {}
input_dict[self.input_tensor.name] = norm_img_batch
outputs = self.predictor.run(self.output_tensors,
input_dict)
preds = outputs[0]
else:
input_names = self.predictor.get_input_names()
for i in range(len(input_names)):
input_tensor = self.predictor.get_input_handle(
input_names[i])
input_tensor.copy_from_cpu(inputs[i])
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.benchmark:
self.autolog.times.stamp()
preds = outputs[0]
else:
if self.use_onnx:
input_dict = {}
input_dict[self.input_tensor.name] = norm_img_batch
outputs = self.predictor.run(self.output_tensors,
input_dict)
preds = outputs[0]
else:
self.input_tensor.copy_from_cpu(norm_img_batch)
self.predictor.run()
outputs = []
for output_tensor in self.output_tensors:
output = output_tensor.copy_to_cpu()
outputs.append(output)
if self.benchmark:
self.autolog.times.stamp()
if len(outputs) != 1:
preds = outputs
else:
preds = outputs[0]
rec_result = self.postprocess_op(preds)
for rno in range(len(rec_result)):
rec_res[indices[beg_img_no + rno]] = rec_result[rno]
......
tools/infer/utility.py
View file @ 237febb1
......@@ -121,6 +121,7 @@ def init_args():
parser.add_argument("--save_log_path", type=str, default="./log_output/")
parser.add_argument("--show_log", type=str2bool, default=True)
parser.add_argument("--use_onnx", type=str2bool, default=False)
return parser
......@@ -144,152 +145,163 @@ def create_predictor(args, mode, logger):
if model_dir is None:
logger.info("not find {} model file path {}".format(mode, model_dir))
sys.exit(0)
model_file_path = model_dir + "/inference.pdmodel"
params_file_path = model_dir + "/inference.pdiparams"
if not os.path.exists(model_file_path):
raise ValueError("not find model file path {}".format(model_file_path))
if not os.path.exists(params_file_path):
raise ValueError("not find params file path {}".format(
params_file_path))
config = inference.Config(model_file_path, params_file_path)
if hasattr(args, 'precision'):
if args.precision == "fp16" and args.use_tensorrt:
precision = inference.PrecisionType.Half
elif args.precision == "int8":
precision = inference.PrecisionType.Int8
else:
precision = inference.PrecisionType.Float32
if args.use_onnx:
import onnxruntime as ort
model_file_path = model_dir
if not os.path.exists(model_file_path):
raise ValueError("not find model file path {}".format(
model_file_path))
sess = ort.InferenceSession(model_file_path)
return sess, sess.get_inputs()[0], None, None
else:
precision = inference.PrecisionType.Float32
if args.use_gpu:
gpu_id = get_infer_gpuid()
if gpu_id is None:
raise ValueError(
"Not found GPU in current device. Please check your device or set args.use_gpu as False"
)
config.enable_use_gpu(args.gpu_mem, 0)
if args.use_tensorrt:
config.enable_tensorrt_engine(
precision_mode=precision,
max_batch_size=args.max_batch_size,
min_subgraph_size=args.min_subgraph_size)
# skip the minmum trt subgraph
if mode == "det":
min_input_shape = {
"x": [1, 3, 50, 50],
"conv2d_92.tmp_0": [1, 120, 20, 20],
"conv2d_91.tmp_0": [1, 24, 10, 10],
"conv2d_59.tmp_0": [1, 96, 20, 20],
"nearest_interp_v2_1.tmp_0": [1, 256, 10, 10],
"nearest_interp_v2_2.tmp_0": [1, 256, 20, 20],
"conv2d_124.tmp_0": [1, 256, 20, 20],
"nearest_interp_v2_3.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_4.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_5.tmp_0": [1, 64, 20, 20],
"elementwise_add_7": [1, 56, 2, 2],
"nearest_interp_v2_0.tmp_0": [1, 256, 2, 2]
}
max_input_shape = {
"x": [1, 3, 1280, 1280],
"conv2d_92.tmp_0": [1, 120, 400, 400],
"conv2d_91.tmp_0": [1, 24, 200, 200],
"conv2d_59.tmp_0": [1, 96, 400, 400],
"nearest_interp_v2_1.tmp_0": [1, 256, 200, 200],
"conv2d_124.tmp_0": [1, 256, 400, 400],
"nearest_interp_v2_2.tmp_0": [1, 256, 400, 400],
"nearest_interp_v2_3.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_4.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_5.tmp_0": [1, 64, 400, 400],
"elementwise_add_7": [1, 56, 400, 400],
"nearest_interp_v2_0.tmp_0": [1, 256, 400, 400]
}
opt_input_shape = {
"x": [1, 3, 640, 640],
"conv2d_92.tmp_0": [1, 120, 160, 160],
"conv2d_91.tmp_0": [1, 24, 80, 80],
"conv2d_59.tmp_0": [1, 96, 160, 160],
"nearest_interp_v2_1.tmp_0": [1, 256, 80, 80],
"nearest_interp_v2_2.tmp_0": [1, 256, 160, 160],
"conv2d_124.tmp_0": [1, 256, 160, 160],
"nearest_interp_v2_3.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_4.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_5.tmp_0": [1, 64, 160, 160],
"elementwise_add_7": [1, 56, 40, 40],
"nearest_interp_v2_0.tmp_0": [1, 256, 40, 40]
}
min_pact_shape = {
"nearest_interp_v2_26.tmp_0": [1, 256, 20, 20],
"nearest_interp_v2_27.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_28.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_29.tmp_0": [1, 64, 20, 20]
}
max_pact_shape = {
"nearest_interp_v2_26.tmp_0": [1, 256, 400, 400],
"nearest_interp_v2_27.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_28.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_29.tmp_0": [1, 64, 400, 400]
}
opt_pact_shape = {
"nearest_interp_v2_26.tmp_0": [1, 256, 160, 160],
"nearest_interp_v2_27.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_28.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_29.tmp_0": [1, 64, 160, 160]
}
min_input_shape.update(min_pact_shape)
max_input_shape.update(max_pact_shape)
opt_input_shape.update(opt_pact_shape)
elif mode == "rec":
min_input_shape = {"x": [1, 3, 32, 10]}
max_input_shape = {"x": [args.rec_batch_num, 3, 32, 1024]}
opt_input_shape = {"x": [args.rec_batch_num, 3, 32, 320]}
elif mode == "cls":
min_input_shape = {"x": [1, 3, 48, 10]}
max_input_shape = {"x": [args.rec_batch_num, 3, 48, 1024]}
opt_input_shape = {"x": [args.rec_batch_num, 3, 48, 320]}
model_file_path = model_dir + "/inference.pdmodel"
params_file_path = model_dir + "/inference.pdiparams"
if not os.path.exists(model_file_path):
raise ValueError("not find model file path {}".format(
model_file_path))
if not os.path.exists(params_file_path):
raise ValueError("not find params file path {}".format(
params_file_path))
config = inference.Config(model_file_path, params_file_path)
if hasattr(args, 'precision'):
if args.precision == "fp16" and args.use_tensorrt:
precision = inference.PrecisionType.Half
elif args.precision == "int8":
precision = inference.PrecisionType.Int8
else:
precision = inference.PrecisionType.Float32
else:
min_input_shape = {"x": [1, 3, 10, 10]}
max_input_shape = {"x": [1, 3, 512, 512]}
opt_input_shape = {"x": [1, 3, 256, 256]}
config.set_trt_dynamic_shape_info(min_input_shape, max_input_shape,
opt_input_shape)
precision = inference.PrecisionType.Float32
if args.use_gpu:
gpu_id = get_infer_gpuid()
if gpu_id is None:
raise ValueError(
"Not found GPU in current device. Please check your device or set args.use_gpu as False"
)
config.enable_use_gpu(args.gpu_mem, 0)
if args.use_tensorrt:
config.enable_tensorrt_engine(
precision_mode=precision,
max_batch_size=args.max_batch_size,
min_subgraph_size=args.min_subgraph_size)
# skip the minmum trt subgraph
if mode == "det":
min_input_shape = {
"x": [1, 3, 50, 50],
"conv2d_92.tmp_0": [1, 120, 20, 20],
"conv2d_91.tmp_0": [1, 24, 10, 10],
"conv2d_59.tmp_0": [1, 96, 20, 20],
"nearest_interp_v2_1.tmp_0": [1, 256, 10, 10],
"nearest_interp_v2_2.tmp_0": [1, 256, 20, 20],
"conv2d_124.tmp_0": [1, 256, 20, 20],
"nearest_interp_v2_3.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_4.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_5.tmp_0": [1, 64, 20, 20],
"elementwise_add_7": [1, 56, 2, 2],
"nearest_interp_v2_0.tmp_0": [1, 256, 2, 2]
}
max_input_shape = {
"x": [1, 3, 1280, 1280],
"conv2d_92.tmp_0": [1, 120, 400, 400],
"conv2d_91.tmp_0": [1, 24, 200, 200],
"conv2d_59.tmp_0": [1, 96, 400, 400],
"nearest_interp_v2_1.tmp_0": [1, 256, 200, 200],
"conv2d_124.tmp_0": [1, 256, 400, 400],
"nearest_interp_v2_2.tmp_0": [1, 256, 400, 400],
"nearest_interp_v2_3.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_4.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_5.tmp_0": [1, 64, 400, 400],
"elementwise_add_7": [1, 56, 400, 400],
"nearest_interp_v2_0.tmp_0": [1, 256, 400, 400]
}
opt_input_shape = {
"x": [1, 3, 640, 640],
"conv2d_92.tmp_0": [1, 120, 160, 160],
"conv2d_91.tmp_0": [1, 24, 80, 80],
"conv2d_59.tmp_0": [1, 96, 160, 160],
"nearest_interp_v2_1.tmp_0": [1, 256, 80, 80],
"nearest_interp_v2_2.tmp_0": [1, 256, 160, 160],
"conv2d_124.tmp_0": [1, 256, 160, 160],
"nearest_interp_v2_3.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_4.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_5.tmp_0": [1, 64, 160, 160],
"elementwise_add_7": [1, 56, 40, 40],
"nearest_interp_v2_0.tmp_0": [1, 256, 40, 40]
}
min_pact_shape = {
"nearest_interp_v2_26.tmp_0": [1, 256, 20, 20],
"nearest_interp_v2_27.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_28.tmp_0": [1, 64, 20, 20],
"nearest_interp_v2_29.tmp_0": [1, 64, 20, 20]
}
max_pact_shape = {
"nearest_interp_v2_26.tmp_0": [1, 256, 400, 400],
"nearest_interp_v2_27.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_28.tmp_0": [1, 64, 400, 400],
"nearest_interp_v2_29.tmp_0": [1, 64, 400, 400]
}
opt_pact_shape = {
"nearest_interp_v2_26.tmp_0": [1, 256, 160, 160],
"nearest_interp_v2_27.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_28.tmp_0": [1, 64, 160, 160],
"nearest_interp_v2_29.tmp_0": [1, 64, 160, 160]
}
min_input_shape.update(min_pact_shape)
max_input_shape.update(max_pact_shape)
opt_input_shape.update(opt_pact_shape)
elif mode == "rec":
min_input_shape = {"x": [1, 3, 32, 10]}
max_input_shape = {"x": [args.rec_batch_num, 3, 32, 1024]}
opt_input_shape = {"x": [args.rec_batch_num, 3, 32, 320]}
elif mode == "cls":
min_input_shape = {"x": [1, 3, 48, 10]}
max_input_shape = {"x": [args.rec_batch_num, 3, 48, 1024]}
opt_input_shape = {"x": [args.rec_batch_num, 3, 48, 320]}
else:
min_input_shape = {"x": [1, 3, 10, 10]}
max_input_shape = {"x": [1, 3, 512, 512]}
opt_input_shape = {"x": [1, 3, 256, 256]}
config.set_trt_dynamic_shape_info(min_input_shape, max_input_shape,
opt_input_shape)
else:
config.disable_gpu()
if hasattr(args, "cpu_threads"):
config.set_cpu_math_library_num_threads(args.cpu_threads)
else:
# default cpu threads as 10
config.set_cpu_math_library_num_threads(10)
if args.enable_mkldnn:
# cache 10 different shapes for mkldnn to avoid memory leak
config.set_mkldnn_cache_capacity(10)
config.enable_mkldnn()
if args.precision == "fp16":
config.enable_mkldnn_bfloat16()
# enable memory optim
config.enable_memory_optim()
config.disable_glog_info()
config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
if mode == 'table':
config.delete_pass("fc_fuse_pass") # not supported for table
config.switch_use_feed_fetch_ops(False)
config.switch_ir_optim(True)
# create predictor
predictor = inference.create_predictor(config)
input_names = predictor.get_input_names()
for name in input_names:
input_tensor = predictor.get_input_handle(name)
output_names = predictor.get_output_names()
output_tensors = []
for output_name in output_names:
output_tensor = predictor.get_output_handle(output_name)
output_tensors.append(output_tensor)
return predictor, input_tensor, output_tensors, config
config.disable_gpu()
if hasattr(args, "cpu_threads"):
config.set_cpu_math_library_num_threads(args.cpu_threads)
else:
# default cpu threads as 10
config.set_cpu_math_library_num_threads(10)
if args.enable_mkldnn:
# cache 10 different shapes for mkldnn to avoid memory leak
config.set_mkldnn_cache_capacity(10)
config.enable_mkldnn()
if args.precision == "fp16":
config.enable_mkldnn_bfloat16()
# enable memory optim
config.enable_memory_optim()
config.disable_glog_info()
config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
if mode == 'table':
config.delete_pass("fc_fuse_pass") # not supported for table
config.switch_use_feed_fetch_ops(False)
config.switch_ir_optim(True)
# create predictor
predictor = inference.create_predictor(config)
input_names = predictor.get_input_names()
for name in input_names:
input_tensor = predictor.get_input_handle(name)
output_names = predictor.get_output_names()
output_tensors = []
for output_name in output_names:
output_tensor = predictor.get_output_handle(output_name)
output_tensors.append(output_tensor)
return predictor, input_tensor, output_tensors, config
def get_infer_gpuid():
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment