Skip to content

P
PaddleOCR

weixin_41840029 / PaddleOCR
与 Fork 源项目一致

Fork自 PaddlePaddle / PaddleOCR

通知 1
Star 1
Fork 0
P
PaddleOCR
提交 038a9330 编写于 作者: A andyjpaddle
浏览文件
操作

Merge branch 'dygraph' of https://github.com/PaddlePaddle/PaddleOCR into dygraph

上级 4805aa0f 91600fcc
隐藏空白更改
内联 并排
Showing with 1878 addition and 231 deletion
PPOCRLabel/libs/shape.py
浏览文件 @ 038a9330
......@@ -48,6 +48,7 @@ class Shape(object):
def __init__(self, label=None, line_color=None, difficult=False, key_cls="None", paintLabel=False):
self.label = label
self.idx = 0
self.points = []
self.fill = False
self.selected = False
......
applications/轻量级车牌识别.md
浏览文件 @ 038a9330
......@@ -311,7 +311,6 @@ python tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml -o
在上述命令中，通过`-o`的方式修改了配置文件中的参数。
训练好的模型地址为: [det_ppocr_v3_finetune.tar](https://paddleocr.bj.bcebos.com/fanliku/license_plate_recognition/det_ppocr_v3_finetune.tar)
**评估**
......@@ -354,8 +353,6 @@ python3.7 deploy/slim/quantization/quant.py -c configs/det/ch_PP-OCRv3/ch_PP-OCR
Eval.dataset.label_file_list=[/home/aistudio/data/CCPD2020/PPOCR/test/det.txt]
```
训练好的模型地址为: [det_ppocr_v3_quant.tar](https://paddleocr.bj.bcebos.com/fanliku/license_plate_recognition/det_ppocr_v3_quant.tar)
量化后指标对比如下
|方案|hmeans| 模型大小 | 预测速度(lite) |
......@@ -436,6 +433,12 @@ python tools/eval.py -c configs/rec/PP-OCRv3/ch_PP-OCRv3_rec.yml -o \
Eval.dataset.label_file_list=[/home/aistudio/data/CCPD2020/PPOCR/test/rec.txt]
```
如需获取已训练模型，请扫码填写问卷，加入PaddleOCR官方交流群获取全部OCR垂类模型下载链接、《动手学OCR》电子书等全套OCR学习资料🎁
<div align="left">
<img src="https://ai-studio-static-online.cdn.bcebos.com/dd721099bd50478f9d5fb13d8dd00fad69c22d6848244fd3a1d3980d7fefc63e" width = "150" height = "150" />
</div>
评估部分日志如下：
```bash
[2022/05/12 19:52:02] ppocr INFO: load pretrain successful from models/ch_PP-OCRv3_rec_train/best_accuracy
......@@ -528,7 +531,6 @@ python tools/train.py -c configs/rec/PP-OCRv3/ch_PP-OCRv3_rec.yml -o \
Eval.dataset.data_dir=/home/aistudio/data/CCPD2020/PPOCR \
Eval.dataset.label_file_list=[/home/aistudio/data/CCPD2020/PPOCR/test/rec.txt]
```
训练好的模型地址为: [rec_ppocr_v3_finetune.tar](https://paddleocr.bj.bcebos.com/fanliku/license_plate_recognition/rec_ppocr_v3_finetune.tar)
**评估**
......@@ -570,7 +572,6 @@ python3.7 deploy/slim/quantization/quant.py -c configs/rec/PP-OCRv3/ch_PP-OCRv3_
Eval.dataset.data_dir=/home/aistudio/data/CCPD2020/PPOCR \
Eval.dataset.label_file_list=[/home/aistudio/data/CCPD2020/PPOCR/test/rec.txt]
```
训练好的模型地址为: [rec_ppocr_v3_quant.tar](https://paddleocr.bj.bcebos.com/fanliku/license_plate_recognition/rec_ppocr_v3_quant.tar)
量化后指标对比如下
......
deploy/avh/.gitignore 0 → 100644
浏览文件 @ 038a9330
include/inputs.h
include/outputs.h
__pycache__/
build/
\ No newline at end of file
deploy/avh/Makefile 0 → 100644
浏览文件 @ 038a9330
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
# Makefile to build demo
# Setup build environment
BUILD_DIR := build
ARM_CPU = ARMCM55
ETHOSU_PATH = /opt/arm/ethosu
CMSIS_PATH ?= ${ETHOSU_PATH}/cmsis
ETHOSU_PLATFORM_PATH ?= ${ETHOSU_PATH}/core_platform
STANDALONE_CRT_PATH := $(abspath $(BUILD_DIR))/runtime
CORSTONE_300_PATH = ${ETHOSU_PLATFORM_PATH}/targets/corstone-300
PKG_COMPILE_OPTS = -g -Wall -O2 -Wno-incompatible-pointer-types -Wno-format -mcpu=cortex-m55 -mthumb -mfloat-abi=hard -std=gnu99
CMAKE ?= cmake
CC = arm-none-eabi-gcc
AR = arm-none-eabi-ar
RANLIB = arm-none-eabi-ranlib
PKG_CFLAGS = ${PKG_COMPILE_OPTS} \
-I${STANDALONE_CRT_PATH}/include \
-I${STANDALONE_CRT_PATH}/src/runtime/crt/include \
-I${PWD}/include \
-I${CORSTONE_300_PATH} \
-I${CMSIS_PATH}/Device/ARM/${ARM_CPU}/Include/ \
-I${CMSIS_PATH}/CMSIS/Core/Include \
-I${CMSIS_PATH}/CMSIS/NN/Include \
-I${CMSIS_PATH}/CMSIS/DSP/Include \
-I$(abspath $(BUILD_DIR))/codegen/host/include
CMSIS_NN_CMAKE_FLAGS = -DCMAKE_TOOLCHAIN_FILE=$(abspath $(BUILD_DIR))/../arm-none-eabi-gcc.cmake \
-DTARGET_CPU=cortex-m55 \
-DBUILD_CMSIS_NN_FUNCTIONS=YES
PKG_LDFLAGS = -lm -specs=nosys.specs -static -T corstone300.ld
$(ifeq VERBOSE,1)
QUIET ?=
$(else)
QUIET ?= @
$(endif)
DEMO_MAIN = src/demo_bare_metal.c
CODEGEN_SRCS = $(wildcard $(abspath $(BUILD_DIR))/codegen/host/src/*.c)
CODEGEN_OBJS = $(subst .c,.o,$(CODEGEN_SRCS))
CMSIS_STARTUP_SRCS = $(wildcard ${CMSIS_PATH}/Device/ARM/${ARM_CPU}/Source/*.c)
UART_SRCS = $(wildcard ${CORSTONE_300_PATH}/*.c)
demo: $(BUILD_DIR)/demo
$(BUILD_DIR)/stack_allocator.o: $(STANDALONE_CRT_PATH)/src/runtime/crt/memory/stack_allocator.c
$(QUIET)mkdir -p $(@D)
$(QUIET)$(CC) -c $(PKG_CFLAGS) -o $@ $^
$(BUILD_DIR)/crt_backend_api.o: $(STANDALONE_CRT_PATH)/src/runtime/crt/common/crt_backend_api.c
$(QUIET)mkdir -p $(@D)
$(QUIET)$(CC) -c $(PKG_CFLAGS) -o $@ $^
# Build generated code
$(BUILD_DIR)/libcodegen.a: $(CODEGEN_SRCS)
$(QUIET)cd $(abspath $(BUILD_DIR)/codegen/host/src) && $(CC) -c $(PKG_CFLAGS) $(CODEGEN_SRCS)
$(QUIET)$(AR) -cr $(abspath $(BUILD_DIR)/libcodegen.a) $(CODEGEN_OBJS)
$(QUIET)$(RANLIB) $(abspath $(BUILD_DIR)/libcodegen.a)
# Build CMSIS startup code
${BUILD_DIR}/libcmsis_startup.a: $(CMSIS_STARTUP_SRCS)
$(QUIET)mkdir -p $(abspath $(BUILD_DIR)/libcmsis_startup)
$(QUIET)cd $(abspath $(BUILD_DIR)/libcmsis_startup) && $(CC) -c $(PKG_CFLAGS) -D${ARM_CPU} $^
$(QUIET)$(AR) -cr $(abspath $(BUILD_DIR)/libcmsis_startup.a) $(abspath $(BUILD_DIR))/libcmsis_startup/*.o
$(QUIET)$(RANLIB) $(abspath $(BUILD_DIR)/libcmsis_startup.a)
CMSIS_SHA_FILE=${CMSIS_PATH}/977abe9849781a2e788b02282986480ff4e25ea6.sha
ifneq ("$(wildcard $(CMSIS_SHA_FILE))","")
${BUILD_DIR}/cmsis_nn/Source/libcmsis-nn.a:
$(QUIET)mkdir -p $(@D)
$(QUIET)cd $(CMSIS_PATH)/CMSIS/NN && $(CMAKE) -B $(abspath $(BUILD_DIR)/cmsis_nn) $(CMSIS_NN_CMAKE_FLAGS)
$(QUIET)cd $(abspath $(BUILD_DIR)/cmsis_nn) && $(MAKE) all
else
# Build CMSIS-NN
${BUILD_DIR}/cmsis_nn/Source/SoftmaxFunctions/libCMSISNNSoftmax.a:
$(QUIET)mkdir -p $(@D)
$(QUIET)cd $(CMSIS_PATH)/CMSIS/NN && $(CMAKE) -B $(abspath $(BUILD_DIR)/cmsis_nn) $(CMSIS_NN_CMAKE_FLAGS)
$(QUIET)cd $(abspath $(BUILD_DIR)/cmsis_nn) && $(MAKE) all
endif
# Build demo application
ifneq ("$(wildcard $(CMSIS_SHA_FILE))","")
$(BUILD_DIR)/demo: $(DEMO_MAIN) $(UART_SRCS) $(BUILD_DIR)/stack_allocator.o $(BUILD_DIR)/crt_backend_api.o \
${BUILD_DIR}/libcodegen.a ${BUILD_DIR}/libcmsis_startup.a ${BUILD_DIR}/cmsis_nn/Source/libcmsis-nn.a
$(QUIET)mkdir -p $(@D)
$(QUIET)$(CC) $(PKG_CFLAGS) $(FREERTOS_FLAGS) -o $@ -Wl,--whole-archive $^ -Wl,--no-whole-archive $(PKG_LDFLAGS)
else
$(BUILD_DIR)/demo: $(DEMO_MAIN) $(UART_SRCS) $(BUILD_DIR)/stack_allocator.o $(BUILD_DIR)/crt_backend_api.o \
${BUILD_DIR}/libcodegen.a ${BUILD_DIR}/libcmsis_startup.a \
${BUILD_DIR}/cmsis_nn/Source/SoftmaxFunctions/libCMSISNNSoftmax.a \
${BUILD_DIR}/cmsis_nn/Source/FullyConnectedFunctions/libCMSISNNFullyConnected.a \
${BUILD_DIR}/cmsis_nn/Source/SVDFunctions/libCMSISNNSVDF.a \
${BUILD_DIR}/cmsis_nn/Source/ReshapeFunctions/libCMSISNNReshape.a \
${BUILD_DIR}/cmsis_nn/Source/ActivationFunctions/libCMSISNNActivation.a \
${BUILD_DIR}/cmsis_nn/Source/NNSupportFunctions/libCMSISNNSupport.a \
${BUILD_DIR}/cmsis_nn/Source/ConcatenationFunctions/libCMSISNNConcatenation.a \
${BUILD_DIR}/cmsis_nn/Source/BasicMathFunctions/libCMSISNNBasicMaths.a \
${BUILD_DIR}/cmsis_nn/Source/ConvolutionFunctions/libCMSISNNConvolutions.a \
${BUILD_DIR}/cmsis_nn/Source/PoolingFunctions/libCMSISNNPooling.a
$(QUIET)mkdir -p $(@D)
$(QUIET)$(CC) $(PKG_CFLAGS) $(FREERTOS_FLAGS) -o $@ -Wl,--whole-archive $^ -Wl,--no-whole-archive $(PKG_LDFLAGS)
endif
clean:
$(QUIET)rm -rf $(BUILD_DIR)/codegen
cleanall:
$(QUIET)rm -rf $(BUILD_DIR)
.SUFFIXES:
.DEFAULT: demo
deploy/avh/README.md 0 → 100644
浏览文件 @ 038a9330
<!--- Licensed to the Apache Software Foundation (ASF) under one -->
<!--- or more contributor license agreements. See the NOTICE file -->
<!--- distributed with this work for additional information -->
<!--- regarding copyright ownership. The ASF licenses this file -->
<!--- to you under the Apache License, Version 2.0 (the -->
<!--- "License"); you may not use this file except in compliance -->
<!--- with the License. You may obtain a copy of the License at -->
<!--- http://www.apache.org/licenses/LICENSE-2.0 -->
<!--- Unless required by applicable law or agreed to in writing, -->
<!--- software distributed under the License is distributed on an -->
<!--- "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY -->
<!--- KIND, either express or implied. See the License for the -->
<!--- specific language governing permissions and limitations -->
<!--- under the License. -->
Running PaddleOCR text recognition model via TVM on bare metal Arm(R) Cortex(R)-M55 CPU and CMSIS-NN
===============================================================
This folder contains an example of how to use TVM to run a PaddleOCR model
on bare metal Cortex(R)-M55 CPU and CMSIS-NN.
Prerequisites
-------------
If the demo is run in the ci_cpu Docker container provided with TVM, then the following
software will already be installed.
If the demo is not run in the ci_cpu Docker container, then you will need the following:
- Software required to build and run the demo (These can all be installed by running
https://github.com/apache/tvm/blob/main/docker/install/ubuntu_install_ethosu_driver_stack.sh .)
- [Fixed Virtual Platform (FVP) based on Arm(R) Corstone(TM)-300 software](https://developer.arm.com/tools-and-software/open-source-software/arm-platforms-software/arm-ecosystem-fvps)
- [cmake 3.19.5](https://github.com/Kitware/CMake/releases/)
- [GCC toolchain from Arm(R)](https://developer.arm.com/-/media/Files/downloads/gnu-rm/10-2020q4/gcc-arm-none-eabi-10-2020-q4-major-x86_64-linux.tar.bz2)
- [Arm(R) Ethos(TM)-U NPU driver stack](https://review.mlplatform.org)
- [CMSIS](https://github.com/ARM-software/CMSIS_5)
- The python libraries listed in the requirements.txt of this directory
- These can be installed by running the following from the current directory:
```bash
pip install -r ./requirements.txt
```
You will also need TVM which can either be:
- Built from source (see [Install from Source](https://tvm.apache.org/docs/install/from_source.html))
- When building from source, the following need to be set in config.cmake:
- set(USE_CMSISNN ON)
- set(USE_MICRO ON)
- set(USE_LLVM ON)
- Installed from TLCPack nightly(see [TLCPack](https://tlcpack.ai/))
You will need to update your PATH environment variable to include the path to cmake 3.19.5 and the FVP.
For example if you've installed these in ```/opt/arm``` , then you would do the following:
```bash
export PATH=/opt/arm/FVP_Corstone_SSE-300/models/Linux64_GCC-6.4:/opt/arm/cmake/bin:$PATH
```
Running the demo application
----------------------------
Type the following command to run the bare metal text recognition application ([src/demo_bare_metal.c](./src/demo_bare_metal.c)):
```bash
./run_demo.sh
```
If the Ethos(TM)-U platform and/or CMSIS have not been installed in /opt/arm/ethosu then
the locations for these can be specified as arguments to run_demo.sh, for example:
```bash
./run_demo.sh --cmsis_path /home/tvm-user/cmsis \
--ethosu_platform_path /home/tvm-user/ethosu/core_platform
```
This will:
- Download a PaddleOCR text recognition model
- Use tvmc to compile the text recognition model for Cortex(R)-M55 CPU and CMSIS-NN
- Create a C header file inputs.c containing the image data as a C array
- Create a C header file outputs.c containing a C array where the output of inference will be stored
- Build the demo application
- Run the demo application on a Fixed Virtual Platform (FVP) based on Arm(R) Corstone(TM)-300 software
- The application will report the text on the image and the corresponding score.
Using your own image
--------------------
The create_image.py script takes a single argument on the command line which is the path of the
image to be converted into an array of bytes for consumption by the model.
The demo can be modified to use an image of your choice by changing the following line in run_demo.sh
```bash
python3 ./convert_image.py path/to/image
```
Model description
-----------------
In this demo, the model we use is an English recognition model based on [PP-OCRv3](https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/doc/doc_ch/PP-OCRv3_introduction.md). PP-OCRv3 is the third version of the PP-OCR series model released by [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR). This series of models has the following features:
- PP-OCRv3: ultra-lightweight OCR system: detection (3.6M) + direction classifier (1.4M) + recognition (12M) = 17.0M
- Support more than 80 kinds of multi-language recognition models, including English, Chinese, French, German, Arabic, Korean, Japanese and so on. For details
- Support vertical text recognition, and long text recognition
The text recognition model in PP-OCRv3 supports more than 80 languages. In the process of model development, since Arm(R) Cortex(R)-M55 CPU does not support rnn operator, we delete the unsupported operator based on the PP-OCRv3 text recognition model to obtain the current model.
\ No newline at end of file
deploy/avh/README_ch.md 0 → 100644
浏览文件 @ 038a9330
<!--- Licensed to the Apache Software Foundation (ASF) under one -->
<!--- or more contributor license agreements. See the NOTICE file -->
<!--- distributed with this work for additional information -->
<!--- regarding copyright ownership. The ASF licenses this file -->
<!--- to you under the Apache License, Version 2.0 (the -->
<!--- "License"); you may not use this file except in compliance -->
<!--- with the License. You may obtain a copy of the License at -->
<!--- http://www.apache.org/licenses/LICENSE-2.0 -->
<!--- Unless required by applicable law or agreed to in writing, -->
<!--- software distributed under the License is distributed on an -->
<!--- "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY -->
<!--- KIND, either express or implied. See the License for the -->
<!--- specific language governing permissions and limitations -->
<!--- under the License. -->
通过TVM在 Arm(R) Cortex(R)-M55 CPU 上运行 PaddleOCR文 本能识别模型
===============================================================
此文件夹包含如何使用 TVM 在 Cortex(R)-M55 CPU 上运行 PaddleOCR 模型的示例。
依赖
-------------
本demo运行在TVM提供的docker环境上，在该环境中已经安装好的必须的软件
在非docker环境中，需要手动安装如下依赖项:
- 软件可通过[安装脚本](https://github.com/apache/tvm/blob/main/docker/install/ubuntu_install_ethosu_driver_stack.sh)一键安装
- [Fixed Virtual Platform (FVP) based on Arm(R) Corstone(TM)-300 software](https://developer.arm.com/tools-and-software/open-source-software/arm-platforms-software/arm-ecosystem-fvps)
- [cmake 3.19.5](https://github.com/Kitware/CMake/releases/)
- [GCC toolchain from Arm(R)](https://developer.arm.com/-/media/Files/downloads/gnu-rm/10-2020q4/gcc-arm-none-eabi-10-2020-q4-major-x86_64-linux.tar.bz2)
- [Arm(R) Ethos(TM)-U NPU driver stack](https://review.mlplatform.org)
- [CMSIS](https://github.com/ARM-software/CMSIS_5)
- python 依赖
```bash
pip install -r ./requirements.txt
```
- TVM
- 从源码安装([Install from Source](https://tvm.apache.org/docs/install/from_source.html))
从源码安装时，需要设置如下字段
- set(USE_CMSISNN ON)
- set(USE_MICRO ON)
- set(USE_LLVM ON)
- 从TLCPack 安装([TLCPack](https://tlcpack.ai/))
安装完成后需要更新环境变量，以软件安装地址为`/opt/arm`为例：
```bash
export PATH=/opt/arm/FVP_Corstone_SSE-300/models/Linux64_GCC-6.4:/opt/arm/cmake/bin:$PATH
```
运行demo
----------------------------
使用如下命令可以一键运行demo
```bash
./run_demo.sh
```
如果 Ethos(TM)-U 平台或 CMSIS 没有安装在 `/opt/arm/ethosu` 中，可通过参数进行设置，例如：
```bash
./run_demo.sh --cmsis_path /home/tvm-user/cmsis \
--ethosu_platform_path /home/tvm-user/ethosu/core_platform
```
`./run_demo.sh`脚本会执行如下步骤:
- 下载 PaddleOCR 文字识别模型
- 使用tvm将PaddleOCR 文字识别模型编译为 Cortex(R)-M55 CPU 和 CMSIS-NN 后端的可执行文件
- 创建一个包含输入图像数据的头文件`inputs.c`
- 创建一个包含输出tensor大小的头文件`outputs.c`
- 编译可执行程序
- 运行程序
- 输出图片上的文字和置信度
使用自己的图片
--------------------
替换 `run_demo.sh ` 中140行处的图片地址即可
使用自己的模型
--------------------
替换 `run_demo.sh ` 中130行处的模型地址即可
模型描述
-----------------
在这个demo中，我们使用的模型是基于[PP-OCRv3](https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/doc/doc_ch/PP-OCRv3_introduction.md)的英文识别模型。 PP-OCRv3是[PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR)发布的PP-OCR系列模型的第三个版本。 该系列模型具有以下特点：
- 超轻量级OCR系统：检测（3.6M）+方向分类器（1.4M）+识别（12M）=17.0M。
- 支持80多种多语言识别模型，包括英文、中文、法文、德文、阿拉伯文、韩文、日文等。
- 支持竖排文本识别，长文本识别。
PP-OCRv3 中的文本识别模型支持 80 多种语言。 在模型开发过程中，由于Arm(R) Cortex(R)-M55 CPU不支持rnn算子，我们在PP-OCRv3文本识别模型的基础上删除了不支持的算子，得到当前模型。
\ No newline at end of file
deploy/avh/arm-none-eabi-gcc.cmake 0 → 100644
浏览文件 @ 038a9330
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
if (__TOOLCHAIN_LOADED)
return()
endif()
set(__TOOLCHAIN_LOADED TRUE)
set(CMAKE_SYSTEM_NAME Generic)
set(CMAKE_C_COMPILER "arm-none-eabi-gcc")
set(CMAKE_CXX_COMPILER "arm-none-eabi-g++")
set(CMAKE_SYSTEM_PROCESSOR "cortex-m55" CACHE STRING "Select Arm(R) Cortex(R)-M architecture. (cortex-m0, cortex-m3, cortex-m33, cortex-m4, cortex-m55, cortex-m7, etc)")
set(CMAKE_TRY_COMPILE_TARGET_TYPE STATIC_LIBRARY)
SET(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
SET(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
SET(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
set(CMAKE_C_STANDARD 99)
set(CMAKE_CXX_STANDARD 14)
# The system processor could for example be set to cortex-m33+nodsp+nofp.
set(__CPU_COMPILE_TARGET ${CMAKE_SYSTEM_PROCESSOR})
string(REPLACE "+" ";" __CPU_FEATURES ${__CPU_COMPILE_TARGET})
list(POP_FRONT __CPU_FEATURES CMAKE_SYSTEM_PROCESSOR)
string(FIND ${__CPU_COMPILE_TARGET} "+" __OFFSET)
if(__OFFSET GREATER_EQUAL 0)
string(SUBSTRING ${__CPU_COMPILE_TARGET} ${__OFFSET} -1 CPU_FEATURES)
endif()
# Add -mcpu to the compile options to override the -mcpu the CMake toolchain adds
add_compile_options(-mcpu=${__CPU_COMPILE_TARGET})
# Set floating point unit
if("${__CPU_COMPILE_TARGET}" MATCHES "\\+fp")
set(FLOAT hard)
elseif("${__CPU_COMPILE_TARGET}" MATCHES "\\+nofp")
set(FLOAT soft)
elseif("${CMAKE_SYSTEM_PROCESSOR}" STREQUAL "cortex-m33" OR
"${CMAKE_SYSTEM_PROCESSOR}" STREQUAL "cortex-m55")
set(FLOAT hard)
else()
set(FLOAT soft)
endif()
add_compile_options(-mfloat-abi=${FLOAT})
add_link_options(-mfloat-abi=${FLOAT})
# Link target
add_link_options(-mcpu=${__CPU_COMPILE_TARGET})
add_link_options(-Xlinker -Map=output.map)
#
# Compile options
#
set(cxx_flags "-fno-unwind-tables;-fno-rtti;-fno-exceptions")
add_compile_options("-Wall;-Wextra;-Wsign-compare;-Wunused;-Wswitch-default;\
-Wdouble-promotion;-Wredundant-decls;-Wshadow;-Wnull-dereference;\
-Wno-format-extra-args;-Wno-unused-function;-Wno-unused-label;\
-Wno-missing-field-initializers;-Wno-return-type;-Wno-format;-Wno-int-conversion"
"$<$<COMPILE_LANGUAGE:CXX>:${cxx_flags}>"
)
deploy/avh/convert_image.py 0 → 100755
浏览文件 @ 038a9330
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
import os
import pathlib
import re
import sys
import cv2
import math
from PIL import Image
import numpy as np
def resize_norm_img(img, image_shape, padding=True):
imgC, imgH, imgW = image_shape
h = img.shape[0]
w = img.shape[1]
if not padding:
resized_image = cv2.resize(
img, (imgW, imgH), interpolation=cv2.INTER_LINEAR)
resized_w = imgW
else:
ratio = w / float(h)
if math.ceil(imgH * ratio) > imgW:
resized_w = imgW
else:
resized_w = int(math.ceil(imgH * ratio))
resized_image = cv2.resize(img, (resized_w, imgH))
resized_image = resized_image.astype('float32')
if image_shape[0] == 1:
resized_image = resized_image / 255
resized_image = resized_image[np.newaxis, :]
else:
resized_image = resized_image.transpose((2, 0, 1)) / 255
resized_image -= 0.5
resized_image /= 0.5
padding_im = np.zeros((imgC, imgH, imgW), dtype=np.float32)
padding_im[:, :, 0:resized_w] = resized_image
return padding_im
def create_header_file(name, tensor_name, tensor_data, output_path):
"""
This function generates a header file containing the data from the numpy array provided.
"""
file_path = pathlib.Path(f"{output_path}/" + name).resolve()
# Create header file with npy_data as a C array
raw_path = file_path.with_suffix(".h").resolve()
with open(raw_path, "w") as header_file:
header_file.write(
"\n"
+ f"const size_t {tensor_name}_len = {tensor_data.size};\n"
+ f'__attribute__((section(".data.tvm"), aligned(16))) float {tensor_name}[] = '
)
header_file.write("{")
for i in np.ndindex(tensor_data.shape):
header_file.write(f"{tensor_data[i]}, ")
header_file.write("};\n\n")
def create_headers(image_name):
"""
This function generates C header files for the input and output arrays required to run inferences
"""
img_path = os.path.join("./", f"{image_name}")
# Resize image to 32x320
img = cv2.imread(img_path)
img = resize_norm_img(img, [3,32,320])
img_data = img.astype("float32")
# # Add the batch dimension, as we are expecting 4-dimensional input: NCHW.
img_data = np.expand_dims(img_data, axis=0)
# Create input header file
create_header_file("inputs", "input", img_data, "./include")
# Create output header file
output_data = np.zeros([7760], np.float)
create_header_file(
"outputs",
"output",
output_data,
"./include",
)
if __name__ == "__main__":
create_headers(sys.argv[1])
deploy/avh/corstone300.ld 0 → 100644
浏览文件 @ 038a9330
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
/*------------------ Reference System Memories -------------
+===================+============+=======+============+============+
| Memory | Address | Size | CPU Access | NPU Access |
+===================+============+=======+============+============+
| ITCM | 0x00000000 | 512KB | Yes (RO) | No |
+-------------------+------------+-------+------------+------------+
| DTCM | 0x20000000 | 512KB | Yes (R/W) | No |
+-------------------+------------+-------+------------+------------+
| SSE-300 SRAM | 0x21000000 | 2MB | Yes (R/W) | Yes (R/W) |
+-------------------+------------+-------+------------+------------+
| Data SRAM | 0x01000000 | 2MB | Yes (R/W) | Yes (R/W) |
+-------------------+------------+-------+------------+------------+
| DDR | 0x60000000 | 32MB | Yes (R/W) | Yes (R/W) |
+-------------------+------------+-------+------------+------------+ */
/*---------------------- ITCM Configuration ----------------------------------
<h> Flash Configuration
<o0> Flash Base Address <0x0-0xFFFFFFFF:8>
<o1> Flash Size (in Bytes) <0x0-0xFFFFFFFF:8>
</h>
-----------------------------------------------------------------------------*/
__ROM_BASE = 0x00000000;
__ROM_SIZE = 0x00080000;
/*--------------------- DTCM RAM Configuration ----------------------------
<h> RAM Configuration
<o0> RAM Base Address <0x0-0xFFFFFFFF:8>
<o1> RAM Size (in Bytes) <0x0-0xFFFFFFFF:8>
</h>
-----------------------------------------------------------------------------*/
__RAM_BASE = 0x20000000;
__RAM_SIZE = 0x00080000;
/*----------------------- Data SRAM Configuration ------------------------------
<h> Data SRAM Configuration
<o0> DATA_SRAM Base Address <0x0-0xFFFFFFFF:8>
<o1> DATA_SRAM Size (in Bytes) <0x0-0xFFFFFFFF:8>
</h>
-----------------------------------------------------------------------------*/
__DATA_SRAM_BASE = 0x01000000;
__DATA_SRAM_SIZE = 0x00200000;
/*--------------------- Embedded SRAM Configuration ----------------------------
<h> SRAM Configuration
<o0> SRAM Base Address <0x0-0xFFFFFFFF:8>
<o1> SRAM Size (in Bytes) <0x0-0xFFFFFFFF:8>
</h>
-----------------------------------------------------------------------------*/
__SRAM_BASE = 0x21000000;
__SRAM_SIZE = 0x00200000;
/*--------------------- Stack / Heap Configuration ----------------------------
<h> Stack / Heap Configuration
<o0> Stack Size (in Bytes) <0x0-0xFFFFFFFF:8>
<o1> Heap Size (in Bytes) <0x0-0xFFFFFFFF:8>
</h>
-----------------------------------------------------------------------------*/
__STACK_SIZE = 0x00008000;
__HEAP_SIZE = 0x00008000;
/*--------------------- Embedded RAM Configuration ----------------------------
<h> DDR Configuration
<o0> DDR Base Address <0x0-0xFFFFFFFF:8>
<o1> DDR Size (in Bytes) <0x0-0xFFFFFFFF:8>
</h>
-----------------------------------------------------------------------------*/
__DDR_BASE = 0x60000000;
__DDR_SIZE = 0x02000000;
/*
*-------------------- <<< end of configuration section >>> -------------------
*/
MEMORY
{
ITCM (rx) : ORIGIN = __ROM_BASE, LENGTH = __ROM_SIZE
DTCM (rwx) : ORIGIN = __RAM_BASE, LENGTH = __RAM_SIZE
DATA_SRAM (rwx) : ORIGIN = __DATA_SRAM_BASE, LENGTH = __DATA_SRAM_SIZE
SRAM (rwx) : ORIGIN = __SRAM_BASE, LENGTH = __SRAM_SIZE
DDR (rwx) : ORIGIN = __DDR_BASE, LENGTH = __DDR_SIZE
}
/* Linker script to place sections and symbol values. Should be used together
* with other linker script that defines memory regions ITCM and RAM.
* It references following symbols, which must be defined in code:
* Reset_Handler : Entry of reset handler
*
* It defines following symbols, which code can use without definition:
* __exidx_start
* __exidx_end
* __copy_table_start__
* __copy_table_end__
* __zero_table_start__
* __zero_table_end__
* __etext
* __data_start__
* __preinit_array_start
* __preinit_array_end
* __init_array_start
* __init_array_end
* __fini_array_start
* __fini_array_end
* __data_end__
* __bss_start__
* __bss_end__
* __end__
* end
* __HeapLimit
* __StackLimit
* __StackTop
* __stack
*/
ENTRY(Reset_Handler)
SECTIONS
{
/* .ddr is placed before .text so that .rodata.tvm is encountered before .rodata* */
.ddr :
{
. = ALIGN (16);
*(.rodata.tvm)
. = ALIGN (16);
*(.data.tvm);
. = ALIGN(16);
} > DDR
.text :
{
KEEP(*(.vectors))
*(.text*)
KEEP(*(.init))
KEEP(*(.fini))
/* .ctors */
*crtbegin.o(.ctors)
*crtbegin?.o(.ctors)
*(EXCLUDE_FILE(*crtend?.o *crtend.o) .ctors)
*(SORT(.ctors.*))
*(.ctors)
/* .dtors */
*crtbegin.o(.dtors)
*crtbegin?.o(.dtors)
*(EXCLUDE_FILE(*crtend?.o *crtend.o) .dtors)
*(SORT(.dtors.*))
*(.dtors)
*(.rodata*)
KEEP(*(.eh_frame*))
} > ITCM
.ARM.extab :
{
*(.ARM.extab* .gnu.linkonce.armextab.*)
} > ITCM
__exidx_start = .;
.ARM.exidx :
{
*(.ARM.exidx* .gnu.linkonce.armexidx.*)
} > ITCM
__exidx_end = .;
.copy.table :
{
. = ALIGN(4);
__copy_table_start__ = .;
LONG (__etext)
LONG (__data_start__)
LONG (__data_end__ - __data_start__)
/* Add each additional data section here */
__copy_table_end__ = .;
} > ITCM
.zero.table :
{
. = ALIGN(4);
__zero_table_start__ = .;
__zero_table_end__ = .;
} > ITCM
/**
* Location counter can end up 2byte aligned with narrow Thumb code but
* __etext is assumed by startup code to be the LMA of a section in DTCM
* which must be 4byte aligned
*/
__etext = ALIGN (4);
.sram :
{
. = ALIGN(16);
} > SRAM AT > SRAM
.data : AT (__etext)
{
__data_start__ = .;
*(vtable)
*(.data)
*(.data.*)
. = ALIGN(4);
/* preinit data */
PROVIDE_HIDDEN (__preinit_array_start = .);
KEEP(*(.preinit_array))
PROVIDE_HIDDEN (__preinit_array_end = .);
. = ALIGN(4);
/* init data */
PROVIDE_HIDDEN (__init_array_start = .);
KEEP(*(SORT(.init_array.*)))
KEEP(*(.init_array))
PROVIDE_HIDDEN (__init_array_end = .);
. = ALIGN(4);
/* finit data */
PROVIDE_HIDDEN (__fini_array_start = .);
KEEP(*(SORT(.fini_array.*)))
KEEP(*(.fini_array))
PROVIDE_HIDDEN (__fini_array_end = .);
KEEP(*(.jcr*))
. = ALIGN(4);
/* All data end */
__data_end__ = .;
} > DTCM
.bss.noinit (NOLOAD):
{
. = ALIGN(16);
*(.bss.noinit.*)
. = ALIGN(16);
} > SRAM AT > SRAM
.bss :
{
. = ALIGN(4);
__bss_start__ = .;
*(.bss)
*(.bss.*)
*(COMMON)
. = ALIGN(4);
__bss_end__ = .;
} > DTCM AT > DTCM
.data_sram :
{
. = ALIGN(16);
} > DATA_SRAM
.heap (COPY) :
{
. = ALIGN(8);
__end__ = .;
PROVIDE(end = .);
. = . + __HEAP_SIZE;
. = ALIGN(8);
__HeapLimit = .;
} > DTCM
.stack (ORIGIN(DTCM) + LENGTH(DTCM) - __STACK_SIZE) (COPY) :
{
. = ALIGN(8);
__StackLimit = .;
. = . + __STACK_SIZE;
. = ALIGN(8);
__StackTop = .;
} > DTCM
PROVIDE(__stack = __StackTop);
/* Check if data + stack exceeds DTCM limit */
ASSERT(__StackLimit >= __bss_end__, "region DTCM overflowed with stack")
}
deploy/avh/include/crt_config.h 0 → 100644
浏览文件 @ 038a9330
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
#ifndef TVM_RUNTIME_CRT_CONFIG_H_
#define TVM_RUNTIME_CRT_CONFIG_H_
/*! Log level of the CRT runtime */
#define TVM_CRT_LOG_LEVEL TVM_CRT_LOG_LEVEL_DEBUG
#endif // TVM_RUNTIME_CRT_CONFIG_H_
deploy/avh/include/tvm_runtime.h 0 → 100644
浏览文件 @ 038a9330
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <tvm/runtime/c_runtime_api.h>
#include <tvm/runtime/crt/stack_allocator.h>
#ifdef __cplusplus
extern "C" {
#endif
void __attribute__((noreturn)) TVMPlatformAbort(tvm_crt_error_t error_code) {
printf("TVMPlatformAbort: %d\n", error_code);
printf("EXITTHESIM\n");
exit(-1);
}
tvm_crt_error_t TVMPlatformMemoryAllocate(size_t num_bytes, DLDevice dev, void** out_ptr) {
return kTvmErrorFunctionCallNotImplemented;
}
tvm_crt_error_t TVMPlatformMemoryFree(void* ptr, DLDevice dev) {
return kTvmErrorFunctionCallNotImplemented;
}
void TVMLogf(const char* msg, ...) {
va_list args;
va_start(args, msg);
vfprintf(stdout, msg, args);
va_end(args);
}
TVM_DLL int TVMFuncRegisterGlobal(const char* name, TVMFunctionHandle f, int override) { return 0; }
#ifdef __cplusplus
}
#endif
deploy/avh/requirements.txt 0 → 100644
浏览文件 @ 038a9330
paddlepaddle
numpy
opencv-python
\ No newline at end of file
deploy/avh/run_demo.sh 0 → 100755
浏览文件 @ 038a9330
#!/bin/bash
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
set -e
set -u
set -o pipefail
# Show usage
function show_usage() {
cat <<EOF
Usage: run_demo.sh
-h, --help
Display this help message.
--cmsis_path CMSIS_PATH
Set path to CMSIS.
--ethosu_platform_path ETHOSU_PLATFORM_PATH
Set path to Arm(R) Ethos(TM)-U core platform.
--fvp_path FVP_PATH
Set path to FVP.
--cmake_path
Set path to cmake.
EOF
}
# Parse arguments
while (( $# )); do
case "$1" in
-h|--help)
show_usage
exit 0
;;
--cmsis_path)
if [ $# -gt 1 ]
then
export CMSIS_PATH="$2"
shift 2
else
echo 'ERROR: --cmsis_path requires a non-empty argument' >&2
show_usage >&2
exit 1
fi
;;
--ethosu_platform_path)
if [ $# -gt 1 ]
then
export ETHOSU_PLATFORM_PATH="$2"
shift 2
else
echo 'ERROR: --ethosu_platform_path requires a non-empty argument' >&2
show_usage >&2
exit 1
fi
;;
--fvp_path)
if [ $# -gt 1 ]
then
export PATH="$2/models/Linux64_GCC-6.4:$PATH"
shift 2
else
echo 'ERROR: --fvp_path requires a non-empty argument' >&2
show_usage >&2
exit 1
fi
;;
--cmake_path)
if [ $# -gt 1 ]
then
export CMAKE="$2"
shift 2
else
echo 'ERROR: --cmake_path requires a non-empty argument' >&2
show_usage >&2
exit 1
fi
;;
-*|--*)
echo "Error: Unknown flag: $1" >&2
show_usage >&2
exit 1
;;
esac
done
# Directories
script_dir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )"
# Make build directory
rm -rf build
make cleanall
mkdir -p build
cd build
wget https://paddleocr.bj.bcebos.com/tvm/ocr_en.tar
tar -xf ocr_en.tar
# Compile model for Arm(R) Cortex(R)-M55 CPU and CMSIS-NN
# An alternative to using "python3 -m tvm.driver.tvmc" is to call
# "tvmc" directly once TVM has been pip installed.
python3 -m tvm.driver.tvmc compile --target=cmsis-nn,c \
--target-cmsis-nn-mcpu=cortex-m55 \
--target-c-mcpu=cortex-m55 \
--runtime=crt \
--executor=aot \
--executor-aot-interface-api=c \
--executor-aot-unpacked-api=1 \
--pass-config tir.usmp.enable=1 \
--pass-config tir.usmp.algorithm=hill_climb \
--pass-config tir.disable_storage_rewrite=1 \
--pass-config tir.disable_vectorize=1 ocr_en/inference.pdmodel \
--output-format=mlf \
--model-format=paddle \
--module-name=rec \
--input-shapes x:[1,3,32,320] \
--output=rec.tar
tar -xf rec.tar
# Create C header files
cd ..
python3 ./convert_image.py imgs_words_en/word_116.png
# Build demo executable
cd ${script_dir}
echo ${script_dir}
make
# Run demo executable on the FVP
FVP_Corstone_SSE-300_Ethos-U55 -C cpu0.CFGDTCMSZ=15 \
-C cpu0.CFGITCMSZ=15 -C mps3_board.uart0.out_file=\"-\" -C mps3_board.uart0.shutdown_tag=\"EXITTHESIM\" \
-C mps3_board.visualisation.disable-visualisation=1 -C mps3_board.telnetterminal0.start_telnet=0 \
-C mps3_board.telnetterminal1.start_telnet=0 -C mps3_board.telnetterminal2.start_telnet=0 -C mps3_board.telnetterminal5.start_telnet=0 \
./build/demo
\ No newline at end of file
deploy/avh/src/demo_bare_metal.c 0 → 100644
浏览文件 @ 038a9330
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
#include <stdio.h>
#include <tvm_runtime.h>
#include <tvmgen_rec.h>
#include "uart.h"
// Header files generated by convert_image.py
#include "inputs.h"
#include "outputs.h"
int main(int argc, char** argv) {
char dict[]={"#0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~!\"#$%&'()*+,-./ "};
int char_dict_nums = 97;
uart_init();
printf("Starting ocr rec inference\n");
struct tvmgen_rec_outputs rec_outputs = {
.output = output,
};
struct tvmgen_rec_inputs rec_inputs = {
.x = input,
};
tvmgen_rec_run(&rec_inputs, &rec_outputs);
// post process
int char_nums = output_len / char_dict_nums;
int last_index = 0;
float score = 0.f;
int count = 0;
printf("text: ");
for (int i = 0; i < char_nums; i++) {
int argmax_idx = 0;
float max_value = 0.0f;
for (int j = 0; j < char_dict_nums; j++){
if (output[i * char_dict_nums + j] > max_value){
max_value = output[i * char_dict_nums + j];
argmax_idx = j;
}
}
if (argmax_idx > 0 && (!(i > 0 && argmax_idx == last_index))) {
score += max_value;
count += 1;
// printf("%d，%f，%c\n", argmax_idx, max_value, dict[argmax_idx]);
printf("%c", dict[argmax_idx]);
}
last_index = argmax_idx;
}
score /= count;
printf(", score: %f\n", score);
// The FVP will shut down when it receives "EXITTHESIM" on the UART
printf("EXITTHESIM\n");
while (1 == 1)
;
return 0;
}
deploy/lite/config.txt
浏览文件 @ 038a9330
......@@ -4,4 +4,5 @@ det_db_box_thresh 0.5
det_db_unclip_ratio 1.6
det_db_use_dilate 0
det_use_polygon_score 1
use_direction_classify 1
\ No newline at end of file
use_direction_classify 1
rec_image_height 32
\ No newline at end of file
deploy/lite/crnn_process.cc
浏览文件 @ 038a9330
......@@ -19,25 +19,27 @@
const std::vector<int> rec_image_shape{3, 32, 320};
cv::Mat CrnnResizeImg(cv::Mat img, float wh_ratio) {
cv::Mat CrnnResizeImg(cv::Mat img, float wh_ratio, int rec_image_height) {
int imgC, imgH, imgW;
imgC = rec_image_shape[0];
imgH = rec_image_height;
imgW = rec_image_shape[2];
imgH = rec_image_shape[1];
imgW = int(32 * wh_ratio);
imgW = int(imgH * wh_ratio);
float ratio = static_cast<float>(img.cols) / static_cast<float>(img.rows);
float ratio = float(img.cols) / float(img.rows);
int resize_w, resize_h;
if (ceilf(imgH * ratio) > imgW)
resize_w = imgW;
else
resize_w = static_cast<int>(ceilf(imgH * ratio));
cv::Mat resize_img;
resize_w = int(ceilf(imgH * ratio));
cv::resize(img, resize_img, cv::Size(resize_w, imgH), 0.f, 0.f,
cv::INTER_LINEAR);
return resize_img;
cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0,
int(imgW - resize_img.cols), cv::BORDER_CONSTANT,
{127, 127, 127});
}
std::vector<std::string> ReadDict(std::string path) {
......
deploy/lite/crnn_process.h
浏览文件 @ 038a9330
......@@ -26,7 +26,7 @@
#include "opencv2/imgcodecs.hpp"
#include "opencv2/imgproc.hpp"
cv::Mat CrnnResizeImg(cv::Mat img, float wh_ratio);
cv::Mat CrnnResizeImg(cv::Mat img, float wh_ratio, int rec_image_height);
std::vector<std::string> ReadDict(std::string path);
......
deploy/lite/ocr_db_crnn.cc
浏览文件 @ 038a9330
......@@ -162,7 +162,8 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
std::vector<std::string> charactor_dict,
std::shared_ptr<PaddlePredictor> predictor_cls,
int use_direction_classify,
std::vector<double> *times) {
std::vector<double> *times,
int rec_image_height) {
std::vector<float> mean = {0.5f, 0.5f, 0.5f};
std::vector<float> scale = {1 / 0.5f, 1 / 0.5f, 1 / 0.5f};
......@@ -183,7 +184,7 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
float wh_ratio =
static_cast<float>(crop_img.cols) / static_cast<float>(crop_img.rows);
resize_img = CrnnResizeImg(crop_img, wh_ratio);
resize_img = CrnnResizeImg(crop_img, wh_ratio, rec_image_height);
resize_img.convertTo(resize_img, CV_32FC3, 1 / 255.f);
const float *dimg = reinterpret_cast<const float *>(resize_img.data);
......@@ -444,7 +445,7 @@ void system(char **argv){
//// load config from txt file
auto Config = LoadConfigTxt(det_config_path);
int use_direction_classify = int(Config["use_direction_classify"]);
int rec_image_height = int(Config["rec_image_height"]);
auto charactor_dict = ReadDict(dict_path);
charactor_dict.insert(charactor_dict.begin(), "#"); // blank char for ctc
charactor_dict.push_back(" ");
......@@ -590,12 +591,16 @@ void rec(int argc, char **argv) {
std::string batchsize = argv[6];
std::string img_dir = argv[7];
std::string dict_path = argv[8];
std::string config_path = argv[9];
if (strcmp(argv[4], "FP32") != 0 && strcmp(argv[4], "INT8") != 0) {
std::cerr << "Only support FP32 or INT8." << std::endl;
exit(1);
}
auto Config = LoadConfigTxt(config_path);
int rec_image_height = int(Config["rec_image_height"]);
std::vector<cv::String> cv_all_img_names;
cv::glob(img_dir, cv_all_img_names);
......@@ -630,7 +635,7 @@ void rec(int argc, char **argv) {
std::vector<float> rec_text_score;
std::vector<double> times;
RunRecModel(boxes, srcimg, rec_predictor, rec_text, rec_text_score,
charactor_dict, cls_predictor, 0, &times);
charactor_dict, cls_predictor, 0, &times, rec_image_height);
//// print recognized text
for (int i = 0; i < rec_text.size(); i++) {
......
deploy/lite/readme.md
浏览文件 @ 038a9330
......@@ -34,7 +34,7 @@ For the compilation process of different development environments, please refer
### 1.2 Prepare Paddle-Lite library
There are two ways to obtain the Paddle-Lite library：
- 1. Download directly, the download link of the Paddle-Lite library is as follows：
- 1. [Recommended] Download directly, the download link of the Paddle-Lite library is as follows：
| Platform | Paddle-Lite library download link |
|---|---|
......@@ -43,7 +43,9 @@ There are two ways to obtain the Paddle-Lite library：
Note: 1. The above Paddle-Lite library is compiled from the Paddle-Lite 2.10 branch. For more information about Paddle-Lite 2.10, please refer to [link](https://github.com/PaddlePaddle/Paddle-Lite/releases/tag/v2.10).
- 2. [Recommended] Compile Paddle-Lite to get the prediction library. The compilation method of Paddle-Lite is as follows：
**Note: It is recommended to use paddlelite>=2.10 version of the prediction library, other prediction library versions [download link](https://github.com/PaddlePaddle/Paddle-Lite/tags)**
- 2. Compile Paddle-Lite to get the prediction library. The compilation method of Paddle-Lite is as follows：
```
git clone https://github.com/PaddlePaddle/Paddle-Lite.git
cd Paddle-Lite
......@@ -104,21 +106,17 @@ If you directly use the model in the above table for deployment, you can skip th
If the model to be deployed is not in the above table, you need to follow the steps below to obtain the optimized model.
The `opt` tool can be obtained by compiling Paddle Lite.
- Step 1: Refer to [document](https://www.paddlepaddle.org.cn/lite/v2.10/user_guides/opt/opt_python.html) to install paddlelite, which is used to convert paddle inference model to paddlelite required for running nb model
```
git clone https://github.com/PaddlePaddle/Paddle-Lite.git
cd Paddle-Lite
git checkout release/v2.10
./lite/tools/build.sh build_optimize_tool
pip install paddlelite==2.10 # The paddlelite version should be the same as the prediction library version
```
After the compilation is complete, the opt file is located under build.opt/lite/api/, You can view the operating options and usage of opt in the following ways:
After installation, the following commands can view the help information
```
cd build.opt/lite/api/
./opt
paddle_lite_opt
```
Introduction to paddle_lite_opt parameters:
|Options|Description|
|---|---|
|--model_dir|The path of the PaddlePaddle model to be optimized (non-combined form)|
......@@ -131,6 +129,8 @@ cd build.opt/lite/api/
`--model_dir` is suitable for the non-combined mode of the model to be optimized, and the inference model of PaddleOCR is the combined mode, that is, the model structure and model parameters are stored in a single file.
- Step 2: Use paddle_lite_opt to convert the inference model to the mobile model format.
The following takes the ultra-lightweight Chinese model of PaddleOCR as an example to introduce the use of the compiled opt file to complete the conversion of the inference model to the Paddle-Lite optimized model
```
......@@ -240,6 +240,7 @@ det_db_thresh 0.3 # Used to filter the binarized image of DB prediction,
det_db_box_thresh 0.5 # DDB post-processing filter box threshold, if there is a missing box detected, it can be reduced as appropriate
det_db_unclip_ratio 1.6 # Indicates the compactness of the text box, the smaller the value, the closer the text box to the text
use_direction_classify 0 # Whether to use the direction classifier, 0 means not to use, 1 means to use
rec_image_height 32 # The height of the input image of the recognition model, the PP-OCRv3 model needs to be set to 48, and the PP-OCRv2 model needs to be set to 32
```
5. Run Model on phone
......@@ -258,8 +259,15 @@ After the above steps are completed, you can use adb to push the file to the pho
cd /data/local/tmp/debug
export LD_LIBRARY_PATH=${PWD}:$LD_LIBRARY_PATH
# The use of ocr_db_crnn is:
# ./ocr_db_crnn Detection model file Orientation classifier model file Recognition model file Test image path Dictionary file path
./ocr_db_crnn ch_PP-OCRv2_det_slim_opt.nb ch_PP-OCRv2_rec_slim_opt.nb ch_ppocr_mobile_v2.0_cls_opt.nb ./11.jpg ppocr_keys_v1.txt
# ./ocr_db_crnn Mode Detection model file Orientation classifier model file Recognition model file Hardware Precision Threads Batchsize Test image path Dictionary file path
./ocr_db_crnn system ch_PP-OCRv2_det_slim_opt.nb ch_PP-OCRv2_rec_slim_opt.nb ch_ppocr_mobile_v2.0_cls_slim_opt.nb arm8 INT8 10 1 ./11.jpg config.txt ppocr_keys_v1.txt True
# precision can be INT8 for quantitative model or FP32 for normal model.
# Only using detection model
./ocr_db_crnn det ch_PP-OCRv2_det_slim_opt.nb arm8 INT8 10 1 ./11.jpg config.txt
# Only using recognition model
./ocr_db_crnn rec ch_PP-OCRv2_rec_slim_opt.nb arm8 INT8 10 1 word_1.jpg ppocr_keys_v1.txt config.txt
```
If you modify the code, you need to recompile and push to the phone.
......@@ -283,3 +291,7 @@ A2: Replace the .jpg test image under ./debug with the image you want to test, a
Q3: How to package it into the mobile APP?
A3: This demo aims to provide the core algorithm part that can run OCR on mobile phones. Further, PaddleOCR/deploy/android_demo is an example of encapsulating this demo into a mobile app for reference.
Q4: When running the demo, an error is reported `Error: This model is not supported, because kernel for 'io_copy' is not supported by Paddle-Lite.`
A4: The problem is that the installed paddlelite version does not match the downloaded prediction library version. Make sure that the paddleliteopt tool matches your prediction library version, and try to switch to the nb model again.
deploy/lite/readme_ch.md
浏览文件 @ 038a9330
......@@ -8,7 +8,7 @@
- [2.1 模型优化](#21-模型优化)
- [2.2 与手机联调](#22-与手机联调)
- [FAQ](#faq)
本教程将介绍基于[Paddle Lite](https://github.com/PaddlePaddle/Paddle-Lite) 在移动端部署PaddleOCR超轻量中文检测、识别模型的详细步骤。
......@@ -32,7 +32,7 @@ Paddle Lite是飞桨轻量化推理引擎，为手机、IOT端提供高效推理
### 1.2 准备预测库
预测库有两种获取方式：
- 1. 直接下载，预测库下载链接如下：
- 1. [推荐]直接下载，预测库下载链接如下：
| 平台 | 预测库下载链接 |
|---|---|
......@@ -41,7 +41,9 @@ Paddle Lite是飞桨轻量化推理引擎，为手机、IOT端提供高效推理
注：1. 上述预测库为PaddleLite 2.10分支编译得到，有关PaddleLite 2.10 详细信息可参考 [链接](https://github.com/PaddlePaddle/Paddle-Lite/releases/tag/v2.10) 。
- 2. [推荐]编译Paddle-Lite得到预测库，Paddle-Lite的编译方式如下：
**注：建议使用paddlelite>=2.10版本的预测库，其他预测库版本[下载链接](https://github.com/PaddlePaddle/Paddle-Lite/tags)**
- 2. 编译Paddle-Lite得到预测库，Paddle-Lite的编译方式如下：
```
git clone https://github.com/PaddlePaddle/Paddle-Lite.git
cd Paddle-Lite
......@@ -102,22 +104,16 @@ Paddle-Lite 提供了多种策略来自动优化原始的模型，其中包括
如果要部署的模型不在上述表格中，则需要按照如下步骤获得优化后的模型。
模型优化需要Paddle-Lite的opt可执行文件，可以通过编译Paddle-Lite源码获得，编译步骤如下：
- 步骤1：参考[文档](https://www.paddlepaddle.org.cn/lite/v2.10/user_guides/opt/opt_python.html)安装paddlelite，用于转换paddle inference model为paddlelite运行所需的nb模型
```
# 如果准备环境时已经clone了Paddle-Lite，则不用重新clone Paddle-Lite
git clone https://github.com/PaddlePaddle/Paddle-Lite.git
cd Paddle-Lite
git checkout release/v2.10
# 启动编译
./lite/tools/build.sh build_optimize_tool
pip install paddlelite==2.10 # paddlelite版本要与预测库版本一致
```
编译完成后，opt文件位于`build.opt/lite/api/`下，可通过如下方式查看opt的运行选项和使用方式；
安装完后，如下指令可以查看帮助信息
```
cd build.opt/lite/api/
./opt
paddle_lite_opt
```
paddle_lite_opt 参数介绍：
|选项|说明|
|---|---|
|--model_dir|待优化的PaddlePaddle模型（非combined形式）的路径|
......@@ -130,6 +126,8 @@ cd build.opt/lite/api/
`--model_dir`适用于待优化的模型是非combined方式，PaddleOCR的inference模型是combined方式，即模型结构和模型参数使用单独一个文件存储。
- 步骤2：使用paddle_lite_opt将inference模型转换成移动端模型格式。
下面以PaddleOCR的超轻量中文模型为例，介绍使用编译好的opt文件完成inference模型到Paddle-Lite优化模型的转换。
```
......@@ -148,7 +146,7 @@ wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/slim/ch_ppocr_mobile_v2.0_cls
转换成功后，inference模型目录下会多出`.nb`结尾的文件，即是转换成功的模型文件。
注意：使用paddle-lite部署时，需要使用opt工具优化后的模型。 opt 工具的输入模型是paddle保存的inference模型
注意：使用paddle-lite部署时，需要使用opt工具优化后的模型。 opt工具的输入模型是paddle保存的inference模型
<a name="2.2与手机联调"></a>
### 2.2 与手机联调
......@@ -234,13 +232,14 @@ ppocr_keys_v1.txt # 中文字典
...
```
2. `config.txt` 包含了检测器、分类器的超参数，如下：
2. `config.txt` 包含了检测器、分类器、识别器的超参数，如下：
```
max_side_len 960 # 输入图像长宽大于960时，等比例缩放图像，使得图像最长边为960
det_db_thresh 0.3 # 用于过滤DB预测的二值化图像，设置为0.-0.3对结果影响不明显
det_db_box_thresh 0.5 # DB后处理过滤box的阈值，如果检测存在漏框情况，可酌情减小
det_db_box_thresh 0.5 # 检测器后处理过滤box的阈值，如果检测存在漏框情况，可酌情减小
det_db_unclip_ratio 1.6 # 表示文本框的紧致程度，越小则文本框更靠近文本
use_direction_classify 0 # 是否使用方向分类器，0表示不使用，1表示使用
rec_image_height 32 # 识别模型输入图像的高度，PP-OCRv3模型设置为48，PP-OCRv2模型需要设置为32
```
5. 启动调试
......@@ -259,8 +258,14 @@ use_direction_classify 0 # 是否使用方向分类器，0表示不使用，1
cd /data/local/tmp/debug
export LD_LIBRARY_PATH=${PWD}:$LD_LIBRARY_PATH
# 开始使用，ocr_db_crnn可执行文件的使用方式为:
# ./ocr_db_crnn 检测模型文件 方向分类器模型文件 识别模型文件 测试图像路径 字典文件路径
./ocr_db_crnn ch_PP-OCRv2_det_slim_opt.nb ch_PP-OCRv2_rec_slim_opt.nb ch_ppocr_mobile_v2.0_cls_slim_opt.nb ./11.jpg ppocr_keys_v1.txt
# ./ocr_db_crnn 预测模式 检测模型文件 方向分类器模型文件 识别模型文件 运行硬件 运行精度 线程数 batchsize 测试图像路径 参数配置路径 字典文件路径 是否使用benchmark参数
./ocr_db_crnn system ch_PP-OCRv2_det_slim_opt.nb ch_PP-OCRv2_rec_slim_opt.nb ch_ppocr_mobile_v2.0_cls_slim_opt.nb arm8 INT8 10 1 ./11.jpg config.txt ppocr_keys_v1.txt True
# 仅使用文本检测模型，使用方式如下：
./ocr_db_crnn det ch_PP-OCRv2_det_slim_opt.nb arm8 INT8 10 1 ./11.jpg config.txt
# 仅使用文本识别模型，使用方式如下：
./ocr_db_crnn rec ch_PP-OCRv2_rec_slim_opt.nb arm8 INT8 10 1 word_1.jpg ppocr_keys_v1.txt config.txt
```
如果对代码做了修改，则需要重新编译并push到手机上。
......@@ -284,3 +289,7 @@ A2：替换debug下的.jpg测试图像为你想要测试的图像，adb push 到
Q3：如何封装到手机APP中？
A3：此demo旨在提供能在手机上运行OCR的核心算法部分，PaddleOCR/deploy/android_demo是将这个demo封装到手机app的示例，供参考
Q4：运行demo时遇到报错`Error: This model is not supported, because kernel for 'io_copy' is not supported by Paddle-Lite.`
A4：问题是安装的paddlelite版本和下载的预测库版本不匹配，确保paddleliteopt工具和你的预测库版本匹配，重新转nb模型试试。
doc/doc_ch/PPOCRv3_det_train.md 0 → 100644
浏览文件 @ 038a9330
# PP-OCRv3 文本检测模型训练
- [1. 简介](#1)
- [2. PPOCRv3检测训练](#2)
- [3. 基于PPOCRv3检测的finetune训练](#3)
<a name="1"></a>
## 1. 简介
PP-OCRv3在PP-OCRv2的基础上进一步升级。本节介绍PP-OCRv3检测模型的训练步骤。有关PPOCRv3策略介绍参考[文档](./PP-OCRv3_introduction.md)。
<a name="2"></a>
## 2. 检测训练
PP-OCRv3检测模型是对PP-OCRv2中的[CML](https://arxiv.org/pdf/2109.03144.pdf)（Collaborative Mutual Learning) 协同互学习文本检测蒸馏策略进行了升级。PP-OCRv3分别针对检测教师模型和学生模型进行进一步效果优化。其中，在对教师模型优化时，提出了大感受野的PAN结构LK-PAN和引入了DML（Deep Mutual Learning）蒸馏策略；在对学生模型优化时，提出了残差注意力机制的FPN结构RSE-FPN。
PP-OCRv3检测训练包括两个步骤：
- 步骤1：采用DML蒸馏方法训练检测教师模型
- 步骤2：使用步骤1得到的教师模型采用CML方法训练出轻量学生模型
### 2.1 准备数据和运行环境
训练数据采用icdar2015数据，准备训练集步骤参考[ocr_dataset](./dataset/ocr_datasets.md).
运行环境准备参考[文档](./installation.md)。
### 2.2 训练教师模型
教师模型训练的配置文件是[ch_PP-OCRv3_det_dml.yml](https://github.com/PaddlePaddle/PaddleOCR/blob/release%2F2.5/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml)。教师模型模型结构的Backbone、Neck、Head分别为Resnet50, LKPAN, DBHead，采用DML的蒸馏方法训练。有关配置文件的详细介绍参考[文档](./knowledge_distillation)。
下载ImageNet预训练模型：
```
# 下载ResNet50_vd的预训练模型
wget -P ./pretrain_models/ https://paddleocr.bj.bcebos.com/pretrained/ResNet50_vd_ssld_pretrained.pdparams
```
**启动训练**
```
# 单卡训练
python3 tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml \
-o Architecture.Models.Student.pretrained=./pretrain_models/ResNet50_vd_ssld_pretrained \
Architecture.Models.Student2.pretrained=./pretrain_models/ResNet50_vd_ssld_pretrained \
Global.save_model_dir=./output/
# 如果要使用多GPU分布式训练，请使用如下命令：
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml \
-o Architecture.Models.Student.pretrained=./pretrain_models/ResNet50_vd_ssld_pretrained \
Architecture.Models.Student2.pretrained=./pretrain_models/ResNet50_vd_ssld_pretrained \
Global.save_model_dir=./output/
```
训练过程中保存的模型在output目录下，包含以下文件：
```
best_accuracy.states
best_accuracy.pdparams # 默认保存最优精度的模型参数
best_accuracy.pdopt # 默认保存最优精度的优化器相关参数
latest.states
latest.pdparams # 默认保存的最新模型参数
latest.pdopt # 默认保存的最新模型的优化器相关参数
```
其中，best_accuracy是保存的精度最高的模型参数，可以直接使用该模型评估。
模型评估命令如下：
```
python3 tools/eval.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml -o Global.checkpoints=./output/best_accuracy
```
训练的教师模型结构更大，精度更高，用于提升学生模型的精度。
**提取教师模型参数**
best_accuracy包含两个模型的参数，分别对应配置文件中的Student，Student2。提取Student的参数方法如下：
```
import paddle
# 加载预训练模型
all_params = paddle.load("output/best_accuracy.pdparams")
# 查看权重参数的keys
print(all_params.keys())
# 模型的权重提取
s_params = {key[len("Student."):]: all_params[key] for key in all_params if "Student." in key}
# 查看模型权重参数的keys
print(s_params.keys())
# 保存
paddle.save(s_params, "./pretrain_models/dml_teacher.pdparams")
```
提取出来的模型参数可以用于模型进一步的finetune训练或者蒸馏训练。
### 2.3 训练学生模型
训练学生模型的配置文件是[ch_PP-OCRv3_det_cml.yml](https://github.com/PaddlePaddle/PaddleOCR/blob/release%2F2.5/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml)
上一节训练得到的教师模型作为监督，采用CML方式训练得到轻量的学生模型。
下载学生模型的ImageNet预训练模型：
```
# 下载MobileNetV3的预训练模型
wget -P ./pretrain_models/ https://paddleocr.bj.bcebos.com/pretrained/MobileNetV3_large_x0_5_pretrained.pdparams
```
**启动训练**
```
# 单卡训练
python3 tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml \
-o Architecture.Models.Student.pretrained=./pretrain_models/MobileNetV3_large_x0_5_pretrained \
Architecture.Models.Student2.pretrained=./pretrain_models/MobileNetV3_large_x0_5_pretrained \
Architecture.Models.Teacher.pretrained=./pretrain_models/dml_teacher \
Global.save_model_dir=./output/
# 如果要使用多GPU分布式训练，请使用如下命令：
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml \
-o Architecture.Models.Student.pretrained=./pretrain_models/MobileNetV3_large_x0_5_pretrained \
Architecture.Models.Student2.pretrained=./pretrain_models/MobileNetV3_large_x0_5_pretrained \
Architecture.Models.Teacher.pretrained=./pretrain_models/dml_teacher \
Global.save_model_dir=./output/
```
训练过程中保存的模型在output目录下，
模型评估命令如下：
```
python3 tools/eval.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml -o Global.checkpoints=./output/best_accuracy
```
best_accuracy包含三个模型的参数，分别对应配置文件中的Student，Student2，Teacher。提取Student参数的方法如下：
```
import paddle
# 加载预训练模型
all_params = paddle.load("output/best_accuracy.pdparams")
# 查看权重参数的keys
print(all_params.keys())
# 模型的权重提取
s_params = {key[len("Student."):]: all_params[key] for key in all_params if "Student." in key}
# 查看模型权重参数的keys
print(s_params.keys())
# 保存
paddle.save(s_params, "./pretrain_models/cml_student.pdparams")
```
提取出来的Student的参数可用于模型部署或者做进一步的finetune训练。
<a name="3"></a>
## 3. 基于PPOCRv3检测finetune训练
本节介绍如何使用PPOCRv3检测模型在其他场景上的finetune训练。
finetune训练适用于三种场景：
- 基于CML蒸馏方法的finetune训练，适用于教师模型在使用场景上精度高于PPOCRv3检测模型，且希望得到一个轻量检测模型。
- 基于PPOCRv3轻量检测模型的finetune训练，无需训练教师模型，希望在PPOCRv3检测模型基础上提升使用场景上的精度。
- 基于DML蒸馏方法的finetune训练，适用于采用DML方法进一步提升精度的场景。
**基于CML蒸馏方法的finetune训练**
下载PPOCRv3训练模型：
```
wget https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_det_distill_train.tar
tar xf ch_PP-OCRv3_det_distill_train.tar
```
ch_PP-OCRv3_det_distill_train/best_accuracy.pdparams包含CML配置文件中Student、Student2、Teacher模型的参数。
启动训练：
```
# 单卡训练
python3 tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml \
-o Global.pretrained_model=./ch_PP-OCRv3_det_distill_train/best_accuracy \
Global.save_model_dir=./output/
# 如果要使用多GPU分布式训练，请使用如下命令：
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml \
-o Global.pretrained_model=./ch_PP-OCRv3_det_distill_train/best_accuracy \
Global.save_model_dir=./output/
```
**基于PPOCRv3轻量检测模型的finetune训练**
下载PPOCRv3训练模型，并提取Student结构的模型参数：
```
wget https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_det_distill_train.tar
tar xf ch_PP-OCRv3_det_distill_train.tar
```
提取Student参数的方法如下：
```
import paddle
# 加载预训练模型
all_params = paddle.load("output/best_accuracy.pdparams")
# 查看权重参数的keys
print(all_params.keys())
# 模型的权重提取
s_params = {key[len("Student."):]: all_params[key] for key in all_params if "Student." in key}
# 查看模型权重参数的keys
print(s_params.keys())
# 保存
paddle.save(s_params, "./student.pdparams")
```
使用配置文件[ch_PP-OCRv3_det_student.yml](https://github.com/PaddlePaddle/PaddleOCR/blob/release%2F2.5/configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml)训练。
**启动训练**
```
# 单卡训练
python3 tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml \
-o Global.pretrained_model=./student \
Global.save_model_dir=./output/
# 如果要使用多GPU分布式训练，请使用如下命令：
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_student.yml \
-o Global.pretrained_model=./student \
Global.save_model_dir=./output/
```
**基于DML蒸馏方法的finetune训练**
以ch_PP-OCRv3_det_distill_train中的Teacher模型为例，首先提取Teacher结构的参数，方法如下：
```
import paddle
# 加载预训练模型
all_params = paddle.load("ch_PP-OCRv3_det_distill_train/best_accuracy.pdparams")
# 查看权重参数的keys
print(all_params.keys())
# 模型的权重提取
s_params = {key[len("Teacher."):]: all_params[key] for key in all_params if "Teacher." in key}
# 查看模型权重参数的keys
print(s_params.keys())
# 保存
paddle.save(s_params, "./teacher.pdparams")
```
**启动训练**
```
# 单卡训练
python3 tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml \
-o Architecture.Models.Student.pretrained=./teacher \
Architecture.Models.Student2.pretrained=./teacher \
Global.save_model_dir=./output/
# 如果要使用多GPU分布式训练，请使用如下命令：
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_dml.yml \
-o Architecture.Models.Student.pretrained=./teacher \
Architecture.Models.Student2.pretrained=./teacher \
Global.save_model_dir=./output/
```
doc/doc_ch/ppocr_introduction.md
浏览文件 @ 038a9330
......@@ -30,11 +30,11 @@ PP-OCR系统pipeline如下：
PP-OCR系统在持续迭代优化，目前已发布PP-OCR和PP-OCRv2两个版本：
PP-OCR从骨干网络选择和调整、预测头部的设计、数据增强、学习率变换策略、正则化参数选择、预训练模型使用以及模型自动裁剪量化8个方面，采用19个有效策略，对各个模块的模型进行效果调优和瘦身(如绿框所示)，最终得到整体大小为3.5M的超轻量中英文OCR和2.8M的英文数字OCR。更多细节请参考PP-OCR技术方案 https://arxiv.org/abs/2009.09941
PP-OCR从骨干网络选择和调整、预测头部的设计、数据增强、学习率变换策略、正则化参数选择、预训练模型使用以及模型自动裁剪量化8个方面，采用19个有效策略，对各个模块的模型进行效果调优和瘦身(如绿框所示)，最终得到整体大小为3.5M的超轻量中英文OCR和2.8M的英文数字OCR。更多细节请参考[PP-OCR技术报告](https://arxiv.org/abs/2009.09941)。
#### PP-OCRv2
PP-OCRv2在PP-OCR的基础上，进一步在5个方面重点优化，检测模型采用CML协同互学习知识蒸馏策略和CopyPaste数据增广策略；识别模型采用LCNet轻量级骨干网络、UDML 改进知识蒸馏策略和[Enhanced CTC loss](./enhanced_ctc_loss.md)损失函数改进（如上图红框所示），进一步在推理速度和预测效果上取得明显提升。更多细节请参考PP-OCRv2[技术报告](https://arxiv.org/abs/2109.03144)。
PP-OCRv2在PP-OCR的基础上，进一步在5个方面重点优化，检测模型采用CML协同互学习知识蒸馏策略和CopyPaste数据增广策略；识别模型采用LCNet轻量级骨干网络、UDML 改进知识蒸馏策略和[Enhanced CTC loss](./enhanced_ctc_loss.md)损失函数改进（如上图红框所示），进一步在推理速度和预测效果上取得明显提升。更多细节请参考[PP-OCRv2技术报告](https://arxiv.org/abs/2109.03144)。
#### PP-OCRv3
......@@ -48,7 +48,7 @@ PP-OCRv3系统pipeline如下：
<img src="../ppocrv3_framework.png" width="800">
</div>
更多细节请参考PP-OCRv3[技术报告](./PP-OCRv3_introduction.md)。
更多细节请参考[PP-OCRv3技术报告](https://arxiv.org/abs/2206.03001v2) 👉[中文简洁版](./PP-OCRv3_introduction.md)
<a name="2"></a>
......
doc/doc_en/ppocr_introduction_en.md
浏览文件 @ 038a9330
......@@ -29,10 +29,10 @@ PP-OCR pipeline is as follows:
PP-OCR system is in continuous optimization. At present, PP-OCR and PP-OCRv2 have been released:
PP-OCR adopts 19 effective strategies from 8 aspects including backbone network selection and adjustment, prediction head design, data augmentation, learning rate transformation strategy, regularization parameter selection, pre-training model use, and automatic model tailoring and quantization to optimize and slim down the models of each module (as shown in the green box above). The final results are an ultra-lightweight Chinese and English OCR model with an overall size of 3.5M and a 2.8M English digital OCR model. For more details, please refer to the PP-OCR technical article (https://arxiv.org/abs/2009.09941).
PP-OCR adopts 19 effective strategies from 8 aspects including backbone network selection and adjustment, prediction head design, data augmentation, learning rate transformation strategy, regularization parameter selection, pre-training model use, and automatic model tailoring and quantization to optimize and slim down the models of each module (as shown in the green box above). The final results are an ultra-lightweight Chinese and English OCR model with an overall size of 3.5M and a 2.8M English digital OCR model. For more details, please refer to [PP-OCR technical report](https://arxiv.org/abs/2009.09941).
#### PP-OCRv2
On the basis of PP-OCR, PP-OCRv2 is further optimized in five aspects. The detection model adopts CML(Collaborative Mutual Learning) knowledge distillation strategy and CopyPaste data expansion strategy. The recognition model adopts LCNet lightweight backbone network, U-DML knowledge distillation strategy and enhanced CTC loss function improvement (as shown in the red box above), which further improves the inference speed and prediction effect. For more details, please refer to the technical report of PP-OCRv2 (https://arxiv.org/abs/2109.03144).
On the basis of PP-OCR, PP-OCRv2 is further optimized in five aspects. The detection model adopts CML(Collaborative Mutual Learning) knowledge distillation strategy and CopyPaste data expansion strategy. The recognition model adopts LCNet lightweight backbone network, U-DML knowledge distillation strategy and enhanced CTC loss function improvement (as shown in the red box above), which further improves the inference speed and prediction effect. For more details, please refer to [PP-OCRv2 technical report](https://arxiv.org/abs/2109.03144).
#### PP-OCRv3
......@@ -46,7 +46,7 @@ PP-OCRv3 pipeline is as follows:
<img src="../ppocrv3_framework.png" width="800">
</div>
For more details, please refer to [PP-OCRv3 technical report](./PP-OCRv3_introduction_en.md).
For more details, please refer to [PP-OCRv3 technical report](https://arxiv.org/abs/2206.03001v2).
<a name="2"></a>
## 2. Features
......
ppocr/data/imaug/fce_targets.py
浏览文件 @ 038a9330
......@@ -107,17 +107,20 @@ class FCENetTargets:
for i in range(1, n):
current_line_len = i * delta_length
while current_line_len >= length_cumsum[current_edge_ind + 1]:
while current_edge_ind + 1 < len(length_cumsum) and current_line_len >= length_cumsum[current_edge_ind + 1]:
current_edge_ind += 1
current_edge_end_shift = current_line_len - length_cumsum[
current_edge_ind]
if current_edge_ind >= len(length_list):
break
end_shift_ratio = current_edge_end_shift / length_list[
current_edge_ind]
current_point = line[current_edge_ind] + (line[current_edge_ind + 1]
- line[current_edge_ind]
) * end_shift_ratio
resampled_line.append(current_point)
resampled_line.append(line[-1])
resampled_line = np.array(resampled_line)
......@@ -328,6 +331,8 @@ class FCENetTargets:
resampled_top_line, resampled_bot_line = self.resample_sidelines(
top_line, bot_line, self.resample_step)
resampled_bot_line = resampled_bot_line[::-1]
if len(resampled_top_line) != len(resampled_bot_line):
continue
center_line = (resampled_top_line + resampled_bot_line) / 2
line_head_shrink_len = norm(resampled_top_line[0] -
......
ppocr/data/imaug/label_ops.py
浏览文件 @ 038a9330
......@@ -23,7 +23,6 @@ import string
from shapely.geometry import LineString, Point, Polygon
import json
import copy
from ppocr.utils.logging import get_logger
......@@ -74,9 +73,10 @@ class DetLabelEncode(object):
s = pts.sum(axis=1)
rect[0] = pts[np.argmin(s)]
rect[2] = pts[np.argmax(s)]
diff = np.diff(pts, axis=1)
rect[1] = pts[np.argmin(diff)]
rect[3] = pts[np.argmax(diff)]
tmp = np.delete(pts, (np.argmin(s), np.argmax(s)), axis=0)
diff = np.diff(np.array(tmp), axis=1)
rect[1] = tmp[np.argmin(diff)]
rect[3] = tmp[np.argmax(diff)]
return rect
def expand_points_num(self, boxes):
......@@ -438,12 +438,14 @@ class KieLabelEncode(object):
texts.append(ann['transcription'])
text_ind = [self.dict[c] for c in text if c in self.dict]
text_inds.append(text_ind)
if 'label' in anno.keys():
if 'label' in ann.keys():
labels.append(ann['label'])
elif 'key_cls' in anno.keys():
labels.append(anno['key_cls'])
elif 'key_cls' in ann.keys():
labels.append(ann['key_cls'])
else:
raise ValueError("Cannot found 'key_cls' in ann.keys(), please check your training annotation.")
raise ValueError(
"Cannot found 'key_cls' in ann.keys(), please check your training annotation."
)
edges.append(ann.get('edge', 0))
ann_infos = dict(
image=data['image'],
......
ppocr/utils/save_load.py
浏览文件 @ 038a9330
......@@ -177,9 +177,9 @@ def save_model(model,
model.backbone.model.save_pretrained(model_prefix)
metric_prefix = os.path.join(model_prefix, 'metric')
# save metric and config
with open(metric_prefix + '.states', 'wb') as f:
pickle.dump(kwargs, f, protocol=2)
if is_best:
with open(metric_prefix + '.states', 'wb') as f:
pickle.dump(kwargs, f, protocol=2)
logger.info('save best model is to {}'.format(model_prefix))
else:
logger.info("save model in {}".format(model_prefix))
test_tipc/configs/ch_PP-OCRv2_det_PACT/train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt → test_tipc/configs/ch_PP-OCRv2_det/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_PP-OCRv2_det_PACT
model_name:ch_PP-OCRv2_det
python:python3.7
gpu_list:0|0,1
Global.use_gpu:True|True
Global.auto_cast:amp
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:fp32
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=50
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=2|whole_train_whole_infer=4
......@@ -12,9 +12,9 @@ train_model_name:latest
train_infer_img_dir:./train_data/icdar2015/text_localization/ch4_test_images/
null:null
##
trainer:pact_train
norm_train:null
pact_train:deploy/slim/quantization/quant.py -c configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml -o
trainer:norm_train
norm_train:tools/train.py -c configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
......@@ -27,8 +27,8 @@ null:null
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:null
quant_export:deploy/slim/quantization/export_model.py -c configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml -o
norm_export:tools/export_model.py -c configs/det/ch_PP-OCRv2/ch_PP-OCRv2_det_cml.yml -o
quant_export:null
fpgm_export:
distill_export:null
export1:null
......@@ -38,7 +38,7 @@ infer_model:./inference/ch_PP-OCRv2_det_infer/
infer_export:null
infer_quant:False
inference:tools/infer/predict_det.py
--use_gpu:True|False
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1
......
test_tipc/configs/ch_PP-OCRv2_rec_PACT/train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt → test_tipc/configs/ch_PP-OCRv2_rec/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_PP-OCRv2_rec_PACT
model_name:ch_PP-OCRv2_rec
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=50
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:fp32
Global.epoch_num:lite_train_lite_infer=3|whole_train_whole_infer=50
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=16|whole_train_whole_infer=128
Global.pretrained_model:pretrain_models/ch_PP-OCRv2_rec_train/best_accuracy
Global.pretrained_model:null
train_model_name:latest
train_infer_img_dir:./inference/rec_inference
null:null
##
trainer:pact_train
norm_train:null
pact_train:deploy/slim/quantization/quant.py -c test_tipc/configs/ch_PP-OCRv2_rec/ch_PP-OCRv2_rec_distillation.yml -o
trainer:norm_train
norm_train:tools/train.py -c test_tipc/configs/ch_PP-OCRv2_rec/ch_PP-OCRv2_rec_distillation.yml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
......@@ -27,18 +27,18 @@ null:null
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:null
quant_export:deploy/slim/quantization/export_model.py -c test_tipc/configs/ch_PP-OCRv2_rec/ch_PP-OCRv2_rec_distillation.yml -o
fpgm_export: null
norm_export:tools/export_model.py -c test_tipc/configs/ch_PP-OCRv2_rec/ch_PP-OCRv2_rec_distillation.yml -o
quant_export:
fpgm_export:
distill_export:null
export1:null
export2:null
inference_dir:Student
infer_model:./inference/ch_PP-OCRv2_rec_slim_quant_infer
infer_model:./inference/ch_PP-OCRv2_rec_infer
infer_export:null
infer_quant:True
infer_quant:False
inference:tools/infer/predict_rec.py
--use_gpu:True|False
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1|6
......
test_tipc/configs/ch_PP-OCRv3_det_PACT/train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt → test_tipc/configs/ch_PP-OCRv3_det/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_PP-OCRv3_det_PACT
model_name:ch_PP-OCRv3_det
python:python3.7
gpu_list:0|0,1
Global.use_gpu:True|True
Global.auto_cast:amp
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:fp32
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=50
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=2|whole_train_whole_infer=4
......@@ -12,9 +12,9 @@ train_model_name:latest
train_infer_img_dir:./train_data/icdar2015/text_localization/ch4_test_images/
null:null
##
trainer:pact_train
norm_train:null
pact_train:deploy/slim/quantization/quant.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml -o
trainer:norm_train
norm_train:tools/train.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
......@@ -27,8 +27,8 @@ null:null
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:null
quant_export:deploy/slim/quantization/export_model.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml -o
norm_export:tools/export_model.py -c configs/det/ch_PP-OCRv3/ch_PP-OCRv3_det_cml.yml -o
quant_export:null
fpgm_export:
distill_export:null
export1:null
......@@ -38,7 +38,7 @@ infer_model:./inference/ch_PP-OCRv3_det_infer/
infer_export:null
infer_quant:False
inference:tools/infer/predict_det.py
--use_gpu:True|False
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1
......
test_tipc/configs/ch_ppocr_mobile_v2.0_det_PACT/train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt → test_tipc/configs/ch_ppocr_mobile_v2.0_det/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_ppocr_mobile_v2.0_det_PACT
model_name:ch_ppocr_mobile_v2.0_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=2|whole_train_whole_infer=50
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:null
Global.epoch_num:lite_train_lite_infer=100|whole_train_whole_infer=50
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
......@@ -12,9 +12,9 @@ train_model_name:latest
train_infer_img_dir:./train_data/icdar2015/text_localization/ch4_test_images/
null:null
##
trainer:pact_train
norm_train:null
pact_train:deploy/slim/quantization/quant.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
trainer:norm_train
norm_train:tools/train.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.pretrained_model=./pretrain_models/MobileNetV3_large_x0_5_pretrained
pact_train:null
fpgm_train:null
distill_train:null
null:null
......@@ -27,18 +27,18 @@ null:null
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:null
quant_export:deploy/slim/quantization/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
norm_export:tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:null
train_model:./inference/ch_ppocr_mobile_v2.0_det_prune_infer/
infer_export: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
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1
......@@ -50,4 +50,4 @@ null:null
--benchmark:True
null:null
===========================infer_benchmark_params==========================
random_infer_input:[{float32,[3,640,640]}];[{float32,[3,960,960]}]
random_infer_input:[{float32,[3,640,640]}];[{float32,[3,960,960]}]
\ No newline at end of file
test_tipc/configs/ch_ppocr_mobile_v2.0_rec_PACT/train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt → test_tipc/configs/ch_ppocr_mobile_v2.0_rec/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_ppocr_mobile_v2.0_rec_PACT
model_name:ch_ppocr_mobile_v2.0_rec
python:python3.7
gpu_list:0
Global.use_gpu:True|True
Global.auto_cast:amp
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=50
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:null
Global.epoch_num:lite_train_lite_infer=2|whole_train_whole_infer=50
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=128|whole_train_whole_infer=128
Global.checkpoints:null
Global.pretrained_model:null
train_model_name:latest
train_infer_img_dir:./train_data/ic15_data/test/word_1.png
train_infer_img_dir:./inference/rec_inference
null:null
##
trainer:pact_train
norm_train:null
pact_train:deploy/slim/quantization/quant.py -c test_tipc/configs/ch_ppocr_mobile_v2.0_rec_PACT/rec_chinese_lite_train_v2.0.yml -o
trainer:norm_train
norm_train:tools/train.py -c configs/rec/rec_icdar15_train.yml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
===========================eval_params===========================
eval:tools/eval.py -c configs/rec/rec_icdar15_train.yml -o
null:null
##
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:null
quant_export:deploy/slim/quantization/export_model.py -c test_tipc/configs/ch_ppocr_mobile_v2.0_rec_PACT/rec_chinese_lite_train_v2.0.yml -o
norm_export:tools/export_model.py -c configs/rec/rec_icdar15_train.yml -o
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:null
infer_model:./inference/ch_ppocr_mobile_v2.0_rec_slim_infer/
infer_export:null
##
train_model:./inference/ch_ppocr_mobile_v2.0_rec_train/best_accuracy
infer_export:tools/export_model.py -c configs/rec/rec_icdar15_train.yml -o
infer_quant:False
inference:tools/infer/predict_rec.py --rec_char_dict_path=./ppocr/utils/ppocr_keys_v1.txt --rec_image_shape="3,32,100"
--use_gpu:True|False
inference:tools/infer/predict_rec.py
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1|6
......@@ -50,4 +50,4 @@ inference:tools/infer/predict_rec.py --rec_char_dict_path=./ppocr/utils/ppocr_ke
--benchmark:True
null:null
===========================infer_benchmark_params==========================
random_infer_input:[{float32,[3,32,320]}]
random_infer_input:[{float32,[3,32,100]}]
test_tipc/configs/ch_ppocr_server_v2.0_det/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt 0 → 100644
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_ppocr_server_v2.0_det
python:python3.7
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:null
Global.epoch_num:lite_train_lite_infer=2|whole_train_whole_infer=50
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=2|whole_train_lite_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
norm_train:tools/train.py -c test_tipc/configs/ch_ppocr_server_v2.0_det/det_r50_vd_db.yml -o
quant_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:tools/eval.py -c test_tipc/configs/ch_ppocr_server_v2.0_det/det_r50_vd_db.yml -o
null:null
##
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:tools/export_model.py -c test_tipc/configs/ch_ppocr_server_v2.0_det/det_r50_vd_db.yml -o
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
##
train_model:./inference/ch_ppocr_server_v2.0_det_train/best_accuracy
infer_export:tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml -o
infer_quant:False
inference:tools/infer/predict_det.py
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1
--use_tensorrt:False
--precision:fp32
--det_model_dir:
--image_dir:./inference/ch_det_data_50/all-sum-510/
--save_log_path:null
--benchmark:True
null:null
===========================infer_benchmark_params==========================
random_infer_input:[{float32,[3,640,640]}];[{float32,[3,960,960]}]
\ No newline at end of file
test_tipc/configs/ch_ppocr_server_v2.0_rec/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt 0 → 100644
浏览文件 @ 038a9330
===========================train_params===========================
model_name:ch_ppocr_server_v2.0_rec
python:python3.7
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:null
Global.epoch_num:lite_train_lite_infer=5|whole_train_whole_infer=100
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=128|whole_train_whole_infer=128
Global.pretrained_model:null
train_model_name:latest
train_infer_img_dir:./inference/rec_inference
null:null
##
trainer:norm_train
norm_train:tools/train.py -c test_tipc/configs/ch_ppocr_server_v2.0_rec/rec_icdar15_train.yml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:tools/eval.py -c test_tipc/configs/ch_ppocr_server_v2.0_rec/rec_icdar15_train.yml -o
null:null
##
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:tools/export_model.py -c test_tipc/configs/ch_ppocr_server_v2.0_rec/rec_icdar15_train.yml -o
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
##
train_model:./inference/ch_ppocr_server_v2.0_rec_train/best_accuracy
infer_export:tools/export_model.py -c test_tipc/configs/ch_ppocr_server_v2.0_rec/rec_icdar15_train.yml -o
infer_quant:False
inference:tools/infer/predict_rec.py
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1|6
--use_tensorrt:False
--precision:fp32
--rec_model_dir:
--image_dir:./inference/rec_inference
--save_log_path:./test/output/
--benchmark:True
null:null
===========================infer_benchmark_params==========================
random_infer_input:[{float32,[3,32,100]}]
test_tipc/configs/det_r50_vd_sast_icdar15_v2.0/train_infer_python.txt
浏览文件 @ 038a9330
......@@ -7,13 +7,13 @@ Global.auto_cast:null
Global.epoch_num:lite_train_lite_infer=1|whole_train_whole_infer=500
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
Global.pretrained_model:./pretrain_models/det_r50_vd_sast_icdar15_v2.0_train/best_accuracy
train_model_name:latest
train_infer_img_dir:./train_data/icdar2015/text_localization/ch4_test_images/
null:null
##
trainer:norm_train
norm_train:tools/train.py -c test_tipc/configs/det_r50_vd_sast_icdar15_v2.0/det_r50_vd_sast_icdar2015.yml -o Global.pretrained_model=./pretrain_models/ResNet50_vd_ssld_pretrained
norm_train:tools/train.py -c test_tipc/configs/det_r50_vd_sast_icdar15_v2.0/det_r50_vd_sast_icdar2015.yml -o
pact_train:null
fpgm_train:null
distill_train:null
......
test_tipc/configs/en_table_structure_PACT/train_linux_gpu_normal_amp_infer_python_linux_gpu_cpu.txt → test_tipc/configs/en_table_structure/train_linux_gpu_fleet_normal_infer_python_linux_gpu_cpu.txt
浏览文件 @ 038a9330
===========================train_params===========================
model_name:en_table_structure_PACT
model_name:en_table_structure
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=50
gpu_list:192.168.0.1,192.168.0.2;0,1
Global.use_gpu:True
Global.auto_cast:fp32
Global.epoch_num:lite_train_lite_infer=3|whole_train_whole_infer=50
Global.save_model_dir:./output/
Train.loader.batch_size_per_card:lite_train_lite_infer=16|whole_train_whole_infer=128
Global.pretrained_model:./pretrain_models/en_ppocr_mobile_v2.0_table_structure_train/best_accuracy
......@@ -12,9 +12,9 @@ train_model_name:latest
train_infer_img_dir:./ppstructure/docs/table/table.jpg
null:null
##
trainer:pact_train
norm_train:null
pact_train:deploy/slim/quantization/quant.py -c test_tipc/configs/en_table_structure/table_mv3.yml -o
trainer:norm_train
norm_train:tools/train.py -c test_tipc/configs/en_table_structure/table_mv3.yml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
......@@ -27,8 +27,8 @@ null:null
===========================infer_params===========================
Global.save_inference_dir:./output/
Global.checkpoints:
norm_export:null
quant_export:deploy/slim/quantization/export_model.py -c test_tipc/configs/en_table_structure/table_mv3.yml -o
norm_export:tools/export_model.py -c test_tipc/configs/en_table_structure/table_mv3.yml -o
quant_export:
fpgm_export:
distill_export:null
export1:null
......@@ -36,9 +36,9 @@ export2:null
##
infer_model:./inference/en_ppocr_mobile_v2.0_table_structure_infer
infer_export:null
infer_quant:True
infer_quant:False
inference:ppstructure/table/predict_table.py --det_model_dir=./inference/en_ppocr_mobile_v2.0_table_det_infer --rec_model_dir=./inference/en_ppocr_mobile_v2.0_table_rec_infer --rec_char_dict_path=./ppocr/utils/dict/table_dict.txt --table_char_dict_path=./ppocr/utils/dict/table_structure_dict.txt --image_dir=./ppstructure/docs/table/table.jpg --det_limit_side_len=736 --det_limit_type=min --output ./output/table
--use_gpu:True|False
--use_gpu:False
--enable_mkldnn:False
--cpu_threads:6
--rec_batch_num:1
......
test_tipc/prepare.sh
浏览文件 @ 038a9330
......@@ -98,8 +98,10 @@ if [ ${MODE} = "lite_train_lite_infer" ];then
fi
if [ ${model_name} == "det_r50_vd_sast_icdar15_v2.0" ] || [ ${model_name} == "det_r50_vd_sast_totaltext_v2.0" ]; then
wget -nc -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_pretrained.pdparams --no-check-certificate
wget -nc -P ./pretrain_models/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_sast_icdar15_v2.0_train.tar --no-check-certificate
wget -nc -P ./train_data/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/total_text_lite.tar --no-check-certificate
cd ./train_data && tar xf total_text_lite.tar && ln -s total_text_lite total_text && cd ../
cd ./pretrain_models && tar xf det_r50_vd_sast_icdar15_v2.0_train.tar && cd ../
fi
if [ ${model_name} == "det_mv3_db_v2_0" ]; then
wget -nc -P ./inference/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_db_v2.0_train.tar --no-check-certificate
......@@ -352,7 +354,7 @@ elif [ ${MODE} = "whole_infer" ];then
fi
fi
if [ ${MODE} = "klquant_whole_infer" ]; then
if [[ ${model_name} =~ "KL" ]]; then
wget -nc -P ./train_data/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar --no-check-certificate
cd ./train_data/ && tar xf icdar2015_lite.tar && rm -rf ./icdar2015 && ln -s ./icdar2015_lite ./icdar2015 && cd ../
if [ ${model_name} = "ch_ppocr_mobile_v2.0_det_KL" ]; then
......
test_tipc/test_paddle2onnx.sh
浏览文件 @ 038a9330
......@@ -62,7 +62,8 @@ function func_paddle2onnx(){
set_save_model=$(func_set_params "--save_file" "${det_save_file_value}")
set_opset_version=$(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} ${set_enable_onnx_checker}"
trans_det_log="${LOG_PATH}/trans_model_det.log"
trans_model_cmd="${padlle2onnx_cmd} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_save_model} ${set_opset_version} ${set_enable_onnx_checker} > ${trans_det_log} 2>&1 "
eval $trans_model_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${trans_model_cmd}" "${status_log}" "${model_name}"
......@@ -73,7 +74,8 @@ function func_paddle2onnx(){
set_save_model=$(func_set_params "--save_file" "${rec_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} ${set_enable_onnx_checker}"
trans_rec_log="${LOG_PATH}/trans_model_rec.log"
trans_model_cmd="${padlle2onnx_cmd} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_save_model} ${set_opset_version} ${set_enable_onnx_checker} > ${trans_rec_log} 2>&1 "
eval $trans_model_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${trans_model_cmd}" "${status_log}" "${model_name}"
......@@ -85,7 +87,8 @@ function func_paddle2onnx(){
set_save_model=$(func_set_params "--save_file" "${det_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} ${set_enable_onnx_checker}"
trans_det_log="${LOG_PATH}/trans_model_det.log"
trans_model_cmd="${padlle2onnx_cmd} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_save_model} ${set_opset_version} ${set_enable_onnx_checker} > ${trans_det_log} 2>&1 "
eval $trans_model_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${trans_model_cmd}" "${status_log}" "${model_name}"
......@@ -97,7 +100,8 @@ function func_paddle2onnx(){
set_save_model=$(func_set_params "--save_file" "${rec_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} ${set_enable_onnx_checker}"
trans_rec_log="${LOG_PATH}/trans_model_rec.log"
trans_model_cmd="${padlle2onnx_cmd} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_save_model} ${set_opset_version} ${set_enable_onnx_checker} > ${trans_rec_log} 2>&1 "
eval $trans_model_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${trans_model_cmd}" "${status_log}" "${model_name}"
......
test_tipc/test_ptq_inference_python.sh 0 → 100644
浏览文件 @ 038a9330
#!/bin/bash
source test_tipc/common_func.sh
FILENAME=$1
# MODE be one of [''whole_infer']
MODE=$2
IFS=$'\n'
# parser klquant_infer params
dataline=$(awk 'NR==1, NR==17{print}' $FILENAME)
lines=(${dataline})
model_name=$(func_parser_value "${lines[1]}")
python=$(func_parser_value "${lines[2]}")
export_weight=$(func_parser_key "${lines[3]}")
save_infer_key=$(func_parser_key "${lines[4]}")
# parser inference model
infer_model_dir_list=$(func_parser_value "${lines[5]}")
infer_export_list=$(func_parser_value "${lines[6]}")
infer_is_quant=$(func_parser_value "${lines[7]}")
# parser inference
inference_py=$(func_parser_value "${lines[8]}")
use_gpu_key=$(func_parser_key "${lines[9]}")
use_gpu_list=$(func_parser_value "${lines[9]}")
use_mkldnn_key=$(func_parser_key "${lines[10]}")
use_mkldnn_list=$(func_parser_value "${lines[10]}")
cpu_threads_key=$(func_parser_key "${lines[11]}")
cpu_threads_list=$(func_parser_value "${lines[11]}")
batch_size_key=$(func_parser_key "${lines[12]}")
batch_size_list=$(func_parser_value "${lines[12]}")
use_trt_key=$(func_parser_key "${lines[13]}")
use_trt_list=$(func_parser_value "${lines[13]}")
precision_key=$(func_parser_key "${lines[14]}")
precision_list=$(func_parser_value "${lines[14]}")
infer_model_key=$(func_parser_key "${lines[15]}")
image_dir_key=$(func_parser_key "${lines[16]}")
infer_img_dir=$(func_parser_value "${lines[16]}")
save_log_key=$(func_parser_key "${lines[17]}")
save_log_value=$(func_parser_value "${lines[17]}")
benchmark_key=$(func_parser_key "${lines[18]}")
benchmark_value=$(func_parser_value "${lines[18]}")
infer_key1=$(func_parser_key "${lines[19]}")
infer_value1=$(func_parser_value "${lines[19]}")
LOG_PATH="./test_tipc/output/${model_name}/${MODE}"
mkdir -p ${LOG_PATH}
status_log="${LOG_PATH}/results_python.log"
function func_inference(){
IFS='|'
_python=$1
_script=$2
_model_dir=$3
_log_path=$4
_img_dir=$5
_flag_quant=$6
# inference
for use_gpu in ${use_gpu_list[*]}; do
if [ ${use_gpu} = "False" ] || [ ${use_gpu} = "cpu" ]; then
for use_mkldnn in ${use_mkldnn_list[*]}; do
for threads in ${cpu_threads_list[*]}; do
for batch_size in ${batch_size_list[*]}; do
for precision in ${precision_list[*]}; do
if [ ${use_mkldnn} = "False" ] && [ ${precision} = "fp16" ]; then
continue
fi # skip when enable fp16 but disable mkldnn
if [ ${_flag_quant} = "True" ] && [ ${precision} != "int8" ]; then
continue
fi # skip when quant model inference but precision is not int8
set_precision=$(func_set_params "${precision_key}" "${precision}")
_save_log_path="${_log_path}/python_infer_cpu_usemkldnn_${use_mkldnn}_threads_${threads}_precision_${precision}_batchsize_${batch_size}.log"
set_infer_data=$(func_set_params "${image_dir_key}" "${_img_dir}")
set_benchmark=$(func_set_params "${benchmark_key}" "${benchmark_value}")
set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
set_mkldnn=$(func_set_params "${use_mkldnn_key}" "${use_mkldnn}")
set_cpu_threads=$(func_set_params "${cpu_threads_key}" "${threads}")
set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
set_infer_params0=$(func_set_params "${save_log_key}" "${save_log_value}")
set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_mkldnn} ${set_cpu_threads} ${set_model_dir} ${set_batchsize} ${set_infer_params0} ${set_infer_data} ${set_benchmark} ${set_precision} ${set_infer_params1} > ${_save_log_path} 2>&1 "
eval $command
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${command}" "${status_log}" "${model_name}"
done
done
done
done
elif [ ${use_gpu} = "True" ] || [ ${use_gpu} = "gpu" ]; then
for use_trt in ${use_trt_list[*]}; do
for precision in ${precision_list[*]}; do
if [ ${_flag_quant} = "True" ] && [ ${precision} != "int8" ]; then
continue
fi # skip when quant model inference but precision is not int8
for batch_size in ${batch_size_list[*]}; do
_save_log_path="${_log_path}/python_infer_gpu_usetrt_${use_trt}_precision_${precision}_batchsize_${batch_size}.log"
set_infer_data=$(func_set_params "${image_dir_key}" "${_img_dir}")
set_benchmark=$(func_set_params "${benchmark_key}" "${benchmark_value}")
set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
set_tensorrt=$(func_set_params "${use_trt_key}" "${use_trt}")
set_precision=$(func_set_params "${precision_key}" "${precision}")
set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
set_infer_params0=$(func_set_params "${save_log_key}" "${save_log_value}")
set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_tensorrt} ${set_precision} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_infer_params1} ${set_infer_params0} > ${_save_log_path} 2>&1 "
eval $command
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${command}" "${status_log}" "${model_name}"
done
done
done
else
echo "Does not support hardware other than CPU and GPU Currently!"
fi
done
}
if [ ${MODE} = "whole_infer" ]; then
GPUID=$3
if [ ${#GPUID} -le 0 ];then
env=" "
else
env="export CUDA_VISIBLE_DEVICES=${GPUID}"
fi
# set CUDA_VISIBLE_DEVICES
eval $env
export Count=0
IFS="|"
infer_run_exports=(${infer_export_list})
infer_quant_flag=(${infer_is_quant})
for infer_model in ${infer_model_dir_list[*]}; do
# run export
if [ ${infer_run_exports[Count]} != "null" ];then
save_infer_dir="${infer_model}_klquant"
set_export_weight=$(func_set_params "${export_weight}" "${infer_model}")
set_save_infer_key=$(func_set_params "${save_infer_key}" "${save_infer_dir}")
export_log_path="${LOG_PATH}/_export_${Count}.log"
export_cmd="${python} ${infer_run_exports[Count]} ${set_export_weight} ${set_save_infer_key} > ${export_log_path} 2>&1 "
echo ${infer_run_exports[Count]}
echo $export_cmd
eval $export_cmd
status_export=$?
status_check $status_export "${export_cmd}" "${status_log}" "${model_name}"
else
save_infer_dir=${infer_model}
fi
#run inference
is_quant="True"
func_inference "${python}" "${inference_py}" "${save_infer_dir}" "${LOG_PATH}" "${infer_img_dir}" ${is_quant}
Count=$(($Count + 1))
done
fi
test_tipc/test_serving_infer_cpp.sh
浏览文件 @ 038a9330
......@@ -70,7 +70,8 @@ function func_serving(){
set_serving_server=$(func_set_params "--serving_server" "${det_serving_server_value}")
set_serving_client=$(func_set_params "--serving_client" "${det_serving_client_value}")
python_list=(${python_list})
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
trans_det_log="${LOG_PATH}/cpp_trans_model_det.log"
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client} > ${trans_det_log} 2>&1 "
eval $trans_model_cmd
cp "deploy/pdserving/serving_client_conf.prototxt" ${det_serving_client_value}
# trans rec
......@@ -78,7 +79,8 @@ function func_serving(){
set_serving_server=$(func_set_params "--serving_server" "${rec_serving_server_value}")
set_serving_client=$(func_set_params "--serving_client" "${rec_serving_client_value}")
python_list=(${python_list})
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
trans_rec_log="${LOG_PATH}/cpp_trans_model_rec.log"
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client} > ${trans_rec_log} 2>&1 "
eval $trans_model_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${trans_model_cmd}" "${status_log}" "${model_name}"
......@@ -88,22 +90,25 @@ function func_serving(){
# cpp serving
for gpu_id in ${gpu_value[*]}; do
if [ ${gpu_id} = "null" ]; then
web_service_cpp_cmd="${python_list[0]} ${web_service_py} --model ${det_server_value} ${rec_server_value} ${op_key} ${op_value} ${port_key} ${port_value} > serving_log_cpu.log &"
server_log_path="${LOG_PATH}/cpp_server_cpu.log"
web_service_cpp_cmd="${python_list[0]} ${web_service_py} --model ${det_server_value} ${rec_server_value} ${op_key} ${op_value} ${port_key} ${port_value} > ${server_log_path} 2>&1 "
eval $web_service_cpp_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cpp_cmd}" "${status_log}" "${model_name}"
sleep 5s
_save_log_path="${LOG_PATH}/server_infer_cpp_cpu.log"
_save_log_path="${LOG_PATH}/cpp_client_cpu.log"
cpp_client_cmd="${python_list[0]} ${cpp_client_py} ${det_client_value} ${rec_client_value} > ${_save_log_path} 2>&1"
eval $cpp_client_cmd
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${cpp_client_cmd}" "${status_log}" "${model_name}"
ps ux | grep -i ${port_value} | awk '{print $2}' | xargs kill -s 9
else
web_service_cpp_cmd="${python_list[0]} ${web_service_py} --model ${det_server_value} ${rec_server_value} ${op_key} ${op_value} ${port_key} ${port_value} ${gpu_key} ${gpu_id} > serving_log_gpu.log &"
server_log_path="${LOG_PATH}/cpp_server_gpu.log"
web_service_cpp_cmd="${python_list[0]} ${web_service_py} --model ${det_server_value} ${rec_server_value} ${op_key} ${op_value} ${port_key} ${port_value} ${gpu_key} ${gpu_id} > ${server_log_path} 2>&1 "
eval $web_service_cpp_cmd
sleep 5s
_save_log_path="${LOG_PATH}/server_infer_cpp_gpu.log"
_save_log_path="${LOG_PATH}/cpp_client_gpu.log"
cpp_client_cmd="${python_list[0]} ${cpp_client_py} ${det_client_value} ${rec_client_value} > ${_save_log_path} 2>&1"
eval $cpp_client_cmd
last_status=${PIPESTATUS[0]}
......
test_tipc/test_serving_infer_python.sh
浏览文件 @ 038a9330
......@@ -77,14 +77,16 @@ function func_serving(){
set_serving_server=$(func_set_params "--serving_server" "${det_serving_server_value}")
set_serving_client=$(func_set_params "--serving_client" "${det_serving_client_value}")
python_list=(${python_list})
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
trans_det_log="${LOG_PATH}/python_trans_model_det.log"
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client} > ${trans_det_log} 2>&1 "
eval $trans_model_cmd
# trans rec
set_dirname=$(func_set_params "--dirname" "${rec_infer_model_dir_value}")
set_serving_server=$(func_set_params "--serving_server" "${rec_serving_server_value}")
set_serving_client=$(func_set_params "--serving_client" "${rec_serving_client_value}")
python_list=(${python_list})
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
trans_rec_log="${LOG_PATH}/python_trans_model_rec.log"
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client} > ${trans_rec_log} 2>&1 "
eval $trans_model_cmd
elif [[ ${model_name} =~ "det" ]]; then
# trans det
......@@ -92,7 +94,8 @@ function func_serving(){
set_serving_server=$(func_set_params "--serving_server" "${det_serving_server_value}")
set_serving_client=$(func_set_params "--serving_client" "${det_serving_client_value}")
python_list=(${python_list})
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
trans_det_log="${LOG_PATH}/python_trans_model_det.log"
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client} > ${trans_det_log} 2>&1 "
eval $trans_model_cmd
elif [[ ${model_name} =~ "rec" ]]; then
# trans rec
......@@ -100,7 +103,8 @@ function func_serving(){
set_serving_server=$(func_set_params "--serving_server" "${rec_serving_server_value}")
set_serving_client=$(func_set_params "--serving_client" "${rec_serving_client_value}")
python_list=(${python_list})
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
trans_rec_log="${LOG_PATH}/python_trans_model_rec.log"
trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client} > ${trans_rec_log} 2>&1 "
eval $trans_model_cmd
fi
set_image_dir=$(func_set_params "${image_dir_key}" "${image_dir_value}")
......@@ -115,29 +119,30 @@ function func_serving(){
for use_mkldnn in ${web_use_mkldnn_list[*]}; do
for threads in ${web_cpu_threads_list[*]}; do
set_cpu_threads=$(func_set_params "${web_cpu_threads_key}" "${threads}")
server_log_path="${LOG_PATH}/python_server_cpu_usemkldnn_${use_mkldnn}_threads_${threads}.log"
if [ ${model_name} = "ch_PP-OCRv2" ] || [ ${model_name} = "ch_PP-OCRv3" ] || [ ${model_name} = "ch_ppocr_mobile_v2.0" ] || [ ${model_name} = "ch_ppocr_server_v2.0" ]; then
set_det_model_config=$(func_set_params "${det_server_key}" "${det_server_value}")
set_rec_model_config=$(func_set_params "${rec_server_key}" "${rec_server_value}")
web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}="" ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_det_model_config} ${set_rec_model_config} &"
web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}="" ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_det_model_config} ${set_rec_model_config} > ${server_log_path} 2>&1 "
eval $web_service_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cmd}" "${status_log}" "${model_name}"
elif [[ ${model_name} =~ "det" ]]; then
set_det_model_config=$(func_set_params "${det_server_key}" "${det_server_value}")
web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}="" ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_det_model_config} &"
web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}="" ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_det_model_config} > ${server_log_path} 2>&1 "
eval $web_service_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cmd}" "${status_log}" "${model_name}"
elif [[ ${model_name} =~ "rec" ]]; then
set_rec_model_config=$(func_set_params "${rec_server_key}" "${rec_server_value}")
web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}="" ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_rec_model_config} &"
web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}="" ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_rec_model_config} > ${server_log_path} 2>&1 "
eval $web_service_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cmd}" "${status_log}" "${model_name}"
fi
sleep 2s
for pipeline in ${pipeline_py[*]}; do
_save_log_path="${LOG_PATH}/server_infer_cpu_${pipeline%_client*}_usemkldnn_${use_mkldnn}_threads_${threads}_batchsize_1.log"
_save_log_path="${LOG_PATH}/python_client_cpu_${pipeline%_client*}_usemkldnn_${use_mkldnn}_threads_${threads}_batchsize_1.log"
pipeline_cmd="${python} ${pipeline} ${set_image_dir} > ${_save_log_path} 2>&1 "
eval $pipeline_cmd
last_status=${PIPESTATUS[0]}
......@@ -151,6 +156,7 @@ function func_serving(){
elif [ ${use_gpu} = "gpu" ]; then
for use_trt in ${web_use_trt_list[*]}; do
for precision in ${web_precision_list[*]}; do
server_log_path="${LOG_PATH}/python_server_gpu_usetrt_${use_trt}_precision_${precision}.log"
if [[ ${_flag_quant} = "False" ]] && [[ ${precision} =~ "int8" ]]; then
continue
fi
......@@ -168,26 +174,26 @@ function func_serving(){
if [ ${model_name} = "ch_PP-OCRv2" ] || [ ${model_name} = "ch_PP-OCRv3" ] || [ ${model_name} = "ch_ppocr_mobile_v2.0" ] || [ ${model_name} = "ch_ppocr_server_v2.0" ]; then
set_det_model_config=$(func_set_params "${det_server_key}" "${det_server_value}")
set_rec_model_config=$(func_set_params "${rec_server_key}" "${rec_server_value}")
web_service_cmd="${python} ${web_service_py} ${set_tensorrt} ${set_precision} ${set_det_model_config} ${set_rec_model_config} &"
web_service_cmd="${python} ${web_service_py} ${set_tensorrt} ${set_precision} ${set_det_model_config} ${set_rec_model_config} > ${server_log_path} 2>&1 "
eval $web_service_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cmd}" "${status_log}" "${model_name}"
elif [[ ${model_name} =~ "det" ]]; then
set_det_model_config=$(func_set_params "${det_server_key}" "${det_server_value}")
web_service_cmd="${python} ${web_service_py} ${set_tensorrt} ${set_precision} ${set_det_model_config} &"
web_service_cmd="${python} ${web_service_py} ${set_tensorrt} ${set_precision} ${set_det_model_config} > ${server_log_path} 2>&1 "
eval $web_service_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cmd}" "${status_log}" "${model_name}"
elif [[ ${model_name} =~ "rec" ]]; then
set_rec_model_config=$(func_set_params "${rec_server_key}" "${rec_server_value}")
web_service_cmd="${python} ${web_service_py} ${set_tensorrt} ${set_precision} ${set_rec_model_config} &"
web_service_cmd="${python} ${web_service_py} ${set_tensorrt} ${set_precision} ${set_rec_model_config} > ${server_log_path} 2>&1 "
eval $web_service_cmd
last_status=${PIPESTATUS[0]}
status_check $last_status "${web_service_cmd}" "${status_log}" "${model_name}"
fi
sleep 2s
for pipeline in ${pipeline_py[*]}; do
_save_log_path="${LOG_PATH}/server_infer_gpu_${pipeline%_client*}_usetrt_${use_trt}_precision_${precision}_batchsize_1.log"
_save_log_path="${LOG_PATH}/python_client_gpu_${pipeline%_client*}_usetrt_${use_trt}_precision_${precision}_batchsize_1.log"
pipeline_cmd="${python} ${pipeline} ${set_image_dir}> ${_save_log_path} 2>&1"
eval $pipeline_cmd
last_status=${PIPESTATUS[0]}
......
test_tipc/test_train_inference_python.sh
浏览文件 @ 038a9330
......@@ -2,7 +2,7 @@
source test_tipc/common_func.sh
FILENAME=$1
# MODE be one of ['lite_train_lite_infer' 'lite_train_whole_infer' 'whole_train_whole_infer', 'whole_infer', 'klquant_whole_infer']
# MODE be one of ['lite_train_lite_infer' 'lite_train_whole_infer' 'whole_train_whole_infer', 'whole_infer']
MODE=$2
dataline=$(awk 'NR==1, NR==51{print}' $FILENAME)
......@@ -88,43 +88,6 @@ benchmark_value=$(func_parser_value "${lines[49]}")
infer_key1=$(func_parser_key "${lines[50]}")
infer_value1=$(func_parser_value "${lines[50]}")
# parser klquant_infer
if [ ${MODE} = "klquant_whole_infer" ]; then
dataline=$(awk 'NR==1, NR==17{print}' $FILENAME)
lines=(${dataline})
model_name=$(func_parser_value "${lines[1]}")
python=$(func_parser_value "${lines[2]}")
export_weight=$(func_parser_key "${lines[3]}")
save_infer_key=$(func_parser_key "${lines[4]}")
# parser inference model
infer_model_dir_list=$(func_parser_value "${lines[5]}")
infer_export_list=$(func_parser_value "${lines[6]}")
infer_is_quant=$(func_parser_value "${lines[7]}")
# parser inference
inference_py=$(func_parser_value "${lines[8]}")
use_gpu_key=$(func_parser_key "${lines[9]}")
use_gpu_list=$(func_parser_value "${lines[9]}")
use_mkldnn_key=$(func_parser_key "${lines[10]}")
use_mkldnn_list=$(func_parser_value "${lines[10]}")
cpu_threads_key=$(func_parser_key "${lines[11]}")
cpu_threads_list=$(func_parser_value "${lines[11]}")
batch_size_key=$(func_parser_key "${lines[12]}")
batch_size_list=$(func_parser_value "${lines[12]}")
use_trt_key=$(func_parser_key "${lines[13]}")
use_trt_list=$(func_parser_value "${lines[13]}")
precision_key=$(func_parser_key "${lines[14]}")
precision_list=$(func_parser_value "${lines[14]}")
infer_model_key=$(func_parser_key "${lines[15]}")
image_dir_key=$(func_parser_key "${lines[16]}")
infer_img_dir=$(func_parser_value "${lines[16]}")
save_log_key=$(func_parser_key "${lines[17]}")
save_log_value=$(func_parser_value "${lines[17]}")
benchmark_key=$(func_parser_key "${lines[18]}")
benchmark_value=$(func_parser_value "${lines[18]}")
infer_key1=$(func_parser_key "${lines[19]}")
infer_value1=$(func_parser_value "${lines[19]}")
fi
LOG_PATH="./test_tipc/output/${model_name}/${MODE}"
mkdir -p ${LOG_PATH}
status_log="${LOG_PATH}/results_python.log"
......@@ -211,7 +174,7 @@ function func_inference(){
done
}
if [ ${MODE} = "whole_infer" ] || [ ${MODE} = "klquant_whole_infer" ]; then
if [ ${MODE} = "whole_infer" ]; then
GPUID=$3
if [ ${#GPUID} -le 0 ];then
env=" "
......@@ -226,16 +189,12 @@ if [ ${MODE} = "whole_infer" ] || [ ${MODE} = "klquant_whole_infer" ]; then
infer_quant_flag=(${infer_is_quant})
for infer_model in ${infer_model_dir_list[*]}; do
# run export
if [ ${infer_run_exports[Count]} != "null" ];then
if [ ${MODE} = "klquant_whole_infer" ]; then
save_infer_dir="${infer_model}_klquant"
fi
if [ ${MODE} = "whole_infer" ]; then
save_infer_dir="${infer_model}"
fi
if [ ${infer_run_exports[Count]} != "null" ];then
save_infer_dir="${infer_model}"
set_export_weight=$(func_set_params "${export_weight}" "${infer_model}")
set_save_infer_key=$(func_set_params "${save_infer_key}" "${save_infer_dir}")
export_cmd="${python} ${infer_run_exports[Count]} ${set_export_weight} ${set_save_infer_key}"
export_log_path="${LOG_PATH}/_export_${Count}.log"
export_cmd="${python} ${infer_run_exports[Count]} ${set_export_weight} ${set_save_infer_key} > ${export_log_path} 2>&1 "
echo ${infer_run_exports[Count]}
echo $export_cmd
eval $export_cmd
......@@ -246,9 +205,6 @@ if [ ${MODE} = "whole_infer" ] || [ ${MODE} = "klquant_whole_infer" ]; then
fi
#run inference
is_quant=${infer_quant_flag[Count]}
if [ ${MODE} = "klquant_whole_infer" ]; then
is_quant="True"
fi
func_inference "${python}" "${inference_py}" "${save_infer_dir}" "${LOG_PATH}" "${infer_img_dir}" ${is_quant}
Count=$(($Count + 1))
done
......@@ -347,7 +303,8 @@ else
if [ ${eval_py} != "null" ]; then
eval ${env}
set_eval_params1=$(func_set_params "${eval_key1}" "${eval_value1}")
eval_cmd="${python} ${eval_py} ${set_eval_pretrain} ${set_use_gpu} ${set_eval_params1}"
eval_log_path="${LOG_PATH}/${trainer}_gpus_${gpu}_autocast_${autocast}_nodes_${nodes}_eval.log"
eval_cmd="${python} ${eval_py} ${set_eval_pretrain} ${set_use_gpu} ${set_eval_params1} > ${eval_log_path} 2>&1 "
eval $eval_cmd
status_check $? "${eval_cmd}" "${status_log}" "${model_name}"
fi
......@@ -355,9 +312,10 @@ else
if [ ${run_export} != "null" ]; then
# run export model
save_infer_path="${save_log}"
export_log_path="${LOG_PATH}/${trainer}_gpus_${gpu}_autocast_${autocast}_nodes_${nodes}_export.log"
set_export_weight=$(func_set_params "${export_weight}" "${save_log}/${train_model_name}")
set_save_infer_key=$(func_set_params "${save_infer_key}" "${save_infer_path}")
export_cmd="${python} ${run_export} ${set_export_weight} ${set_save_infer_key}"
export_cmd="${python} ${run_export} ${set_export_weight} ${set_save_infer_key} > ${export_log_path} 2>&1 "
eval $export_cmd
status_check $? "${export_cmd}" "${status_log}" "${model_name}"
......
tools/infer/predict_det.py
浏览文件 @ 038a9330
......@@ -154,9 +154,10 @@ class TextDetector(object):
s = pts.sum(axis=1)
rect[0] = pts[np.argmin(s)]
rect[2] = pts[np.argmax(s)]
diff = np.diff(pts, axis=1)
rect[1] = pts[np.argmin(diff)]
rect[3] = pts[np.argmax(diff)]
tmp = np.delete(pts, (np.argmin(s), np.argmax(s)), axis=0)
diff = np.diff(np.array(tmp), axis=1)
rect[1] = tmp[np.argmin(diff)]
rect[3] = tmp[np.argmax(diff)]
return rect
def clip_det_res(self, points, img_height, img_width):
......
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