海思ss928部署手写数字识别模型

大致流程---------------------------------------------------------------------------------------------------------------------

模型转换----------------------------------------------------------------------------------------------------

1：准备MNIST的onnx模型---> https://kdocs.cn/l/ctULnY8mxXuE

github地址--> GitHub - warren-wzw/MNIST-pytorch

搭建好ATC的环境--> https://kdocs.cn/l/cjeQxGjfojsX

首先设置环境变量

source /home/warren/Ascend/ascend-toolkit/latest/x86_64-linux/bin/setenv.bash

转化模型

atc --model=/home/warren/Ascend/yolov5/model/yolov5s.onnx \

--soc_version=OPTG  --framework=5 \

--output=/home/warren/Ascend/yolov5/model/yolov5s \

--input_shape="input0:1,2,64,64"  

atc --model=/home/warren/ss928/NNN_PC/amct/amct_onnx/sample/MNIST/outputs/calibration/MNIST_deploy_model.onnx \

> --soc_version=OPTG  --framework=5 \

> --output=/home/warren/ss928/NNN_PC/amct/amct_onnx/sample/MNIST/outputs/calibration/MNIST \

> --input_shape="input0:1,3,640,640"

模型转化成功后得到以下文件

        

模型量化--------------------------------------------------------------------------------------------------------------

目录结构

import os
import argparse
import cv2
import numpy as np
import onnxruntime as ort
import time
import torchimport amct_onnx as amctPATH = os.path.realpath('./')
DATA_DIR = os.path.join(PATH, 'data')
PARSER = argparse.ArgumentParser(description='amct_onnx MNIST quantization sample.')
ARGS = PARSER.parse_args()
OUTPUTS = os.path.join(PATH, 'outputs/calibration')
TMP = os.path.join(OUTPUTS, 'tmp')def onnx_forward(onnx_model, batch_size=1, iterations=100):ort_session = ort.InferenceSession(onnx_model, amct.AMCT_SO)with open("./data/train-images-idx3-ubyte","rb") as f:file = f.read()num = -1j=0inference_time =[0]for j in range(100):num=num+1i = 16+784*numimage1 = [int(str(item).encode('ascii'),16) for item in file[i:i+784]]input_data = np.array(image1,dtype=np.float32).reshape(1,1,28,28)#np.set_printoptions(linewidth=150)#print(input_data)input_name = ort_session.get_inputs()[0].name# inferencestart_time = time.time()output = ort_session.run(None, {input_name: input_data})end_time = time.time()inference_time.append(end_time - start_time)# 处理输出结果output = torch.tensor(output[0])  # 将输出转换为 PyTorch 张量#print(output_tensor)# 输出结果处理和后续操作...pred =np.argmax(output)print("------------------------The num of this pic is ",pred,"use time ",inference_time[num]*1000,"ms",j)def main():model_file = './model/model.onnx'print('[INFO] Do original model test:')onnx_forward(model_file,1,1)config_json_file = os.path.join(TMP, 'config.json')skip_layers = []amct.create_quant_config(config_file=config_json_file, model_file=model_file, skip_layers=skip_layers, batch_num=1,activation_offset=True, config_defination=None)# Phase1: do conv+bn fusion, weights calibration and generate#         calibration modelscale_offset_record_file = os.path.join(TMP, 'record.txt')modified_model = os.path.join(TMP, 'modified_model.onnx')amct.quantize_model(config_file=config_json_file, model_file=model_file, modified_onnx_file=modified_model,record_file=scale_offset_record_file)onnx_forward(modified_model, 32, 1)# Phase3: save final model, one for onnx do fake quant test, one#         deploy model for ATCresult_path = os.path.join(OUTPUTS, 'MNIST')amct.save_model(modified_model, scale_offset_record_file, result_path)# Phase4: run fake_quant model testprint('[INFO] Do quantized model test:')onnx_forward('%s_%s' % (result_path, 'fake_quant_model.onnx'), 1, 1)
if __name__ == '__main__':main()

推理代码编写---------------------------------------------------------------------------------------------------

将官方的sample复制一份改为MNIST，目录结构如下图所示

更改camke文件

1：添加环境变量：

export DDK_PATH=$HOME/Ascend/ascend-toolkit/latest

export NPU_HOST_LIB=$DDK_PATH/runtime/lib64/stub

2：创建build目录

mkdir -p build/intermediates/host

3：cmake ../../../src -DCMAKE_CXX_COMPILER=aarch64-mix210-linux-g++ -DCMAKE_SKIP_RPATH=TRUE

将整个MNIST文件夹拷贝至板端，添加库文件路径的环境变量

export ASCEND_GLOBAL_EVENT_ENABLE=0

export ASCEND_AACPU_KERNEL_PATH=/opt/sd/lib

export ASCEND_AICPU_KERNEL_PATH=/opt/sd/lib

export LD_LIBRARY_PATH=/opt/sd/lib

执行可执行文件main

执行成功。

代码讲解-------------------------------------------------------------------------

大致逻辑

 

main.cpp

#include "main.h"
#include "acl/acl.h"#define INFO_LOG(fmt, ...)  fprintf(stdout, "[INFO]  " fmt "\n", ##__VA_ARGS__)
#define WARN_LOG(fmt, ...)  fprintf(stdout, "[WARN]  " fmt "\n", ##__VA_ARGS__)
#define ERROR_LOG(fmt, ...) fprintf(stderr, "[ERROR]  " fmt "\n", ##__VA_ARGS__)
const int MODEL_CHANNEL = 1;
const int MODEL_IN_WIDTH = 28;
const int MODEL_IN_HEIGHT = 28;
const int loop_count = 1000;typedef enum Result {SUCCESS = 0,FAILED = 1
} Result;static inline int64_t getCurrentTimeUs()
{struct timeval tv;gettimeofday(&tv, NULL);return tv.tv_sec * 1000000 + tv.tv_usec;
}
void Load_data(int num,unsigned char * input_image)
{int j=16+784*num;FILE *file = fopen("../data/train-images-idx3-ubyte", "rb");if (file == NULL) {printf("can't open the file!\n");}fseek(file,j,SEEK_SET);fread(input_image,sizeof(char),784,file);//print
/*     for(int i=0;i<MODEL_IN_WIDTH;i++){for(int j=0;j<MODEL_IN_WIDTH;j++){printf("%4d",input_image[i*28+j]);}printf("\n");}  */fclose(file);
}
void Bubble_sort(float *buffer,int num)
{float temp;for(int i=0; i<num;i++){for(int j=0; j<num-i-1;j++){if(buffer[j]>buffer[j+1]){temp = buffer[j];buffer[j]=buffer[j+1];buffer[j+1]=temp;}}}
}
int main()
{
/***************************************************/
/*****************define var************************/
/***************************************************/int num=0;aclError ret=1;const char *aclConfigPath = "../src/acl.json";int32_t deviceId_=0;aclrtContext context_=nullptr;aclrtStream stream_=nullptr;aclrtRunMode runMode;uint32_t modelId_=0;const char* modelPath = "../model/MNIST.om";aclmdlDesc *modelDesc_;aclmdlDataset *output_;aclmdlDataset *input_;void * inputDataBuffer = nullptr;size_t size = 784;void* input_image_original;void* time_ori;int64_t sum=0;int64_t start_time=0;int64_t end_time=0;
/***************************************************/
/*****************Init ACL**************************/
/***************************************************/ret = aclInit(aclConfigPath);if (ret != ACL_SUCCESS) {ERROR_LOG("acl init failed, errorCode = %d", static_cast<int32_t>(ret));return FAILED;}INFO_LOG("--------------------acl init success");
/***************************************************/
/*****************apply resource********************/
/***************************************************/// set device only one device    ret = aclrtSetDevice(deviceId_);if (ret != ACL_SUCCESS) {ERROR_LOG("acl set device %d failed, errorCode = %d", deviceId_, static_cast<int32_t>(ret));return FAILED;}INFO_LOG("set device %d success", deviceId_);// create context (set current)ret = aclrtCreateContext(&context_, deviceId_);if (ret != ACL_SUCCESS) {ERROR_LOG("acl create context failed, deviceId = %d, errorCode = %d",deviceId_, static_cast<int32_t>(ret));return FAILED;}INFO_LOG("create context success");// create streamret = aclrtCreateStream(&stream_);if (ret != ACL_SUCCESS) {ERROR_LOG("acl create stream failed, deviceId = %d, errorCode = %d",deviceId_, static_cast<int32_t>(ret));return FAILED;}INFO_LOG("create stream success");// get run moderet = aclrtGetRunMode(&runMode);if (ret != ACL_SUCCESS) {ERROR_LOG("acl get run mode failed, errorCode = %d", static_cast<int32_t>(ret));return FAILED;} 
/***************************************************/
/********load model and get infos of model**********/
/***************************************************/ret = aclmdlLoadFromFile(modelPath,&modelId_);if (ret != ACL_SUCCESS) {ERROR_LOG("load model from file failed, model file is %s, errorCode is %d",modelPath, static_cast<int32_t>(ret));return FAILED;}INFO_LOG("load model %s success id is %d\n", modelPath,modelId_);//get model describemodelDesc_ = aclmdlCreateDesc();if (modelDesc_ == nullptr) {ERROR_LOG("create model description failed");return FAILED;}ret = aclmdlGetDesc(modelDesc_, modelId_);if (ret != ACL_SUCCESS) {ERROR_LOG("get model description failed, modelId is %u, errorCode is %d",modelId_, static_cast<int32_t>(ret));return FAILED;}INFO_LOG("create model description success");
/***************************************************/
/******************print input tensor***************/
/***************************************************/     
/*     aclmdlIODims *dim;ret=aclmdlGetInputDims(modelDesc_,0,dim);printf("----------------in dims is %d \n",dim->dimCount);printf("----------------in dims name is: %s dims: \n",dim->name);for(int num=0;num<dim->dimCount;num++){printf("%d ",num,dim->dims[num]);}ret = aclmdlGetOutputDims(modelDesc_,0,dim);printf("----------------out dims is %d \n",dim->dimCount);printf("----------------out dims name is: %s dims:\n",dim->name);for(int num=0;num<dim->dimCount;num++){printf("%d \n",num,dim->dims[num]);} deviceId_=0;*/
/***************************************************/
/******************prepare output data buffer***************/
/***************************************************/output_ = aclmdlCreateDataset();if (output_ == nullptr) {ERROR_LOG("can't create dataset, create output failed");return FAILED;}size_t outputSize = aclmdlGetNumOutputs(modelDesc_); for (size_t i = 0; i < outputSize; ++i) {size_t modelOutputSize = aclmdlGetOutputSizeByIndex(modelDesc_, i);void *outputBuffer = nullptr;ret = aclrtMalloc(&outputBuffer, modelOutputSize, ACL_MEM_MALLOC_NORMAL_ONLY);if (ret != ACL_SUCCESS) {ERROR_LOG("can't malloc buffer, size is %zu, create output failed, errorCode is %d",modelOutputSize, static_cast<int32_t>(ret));return FAILED;}//apply output bufferaclDataBuffer *outputData = aclCreateDataBuffer(outputBuffer, modelOutputSize);if (outputData == nullptr) {ERROR_LOG("can't create data buffer, create output failed");(void)aclrtFree(outputBuffer);return FAILED;}ret = aclmdlAddDatasetBuffer(output_, outputData);if (ret != ACL_SUCCESS) {ERROR_LOG("can't add data buffer, create output failed, errorCode is %d",static_cast<int32_t>(ret));(void)aclrtFree(outputBuffer);(void)aclDestroyDataBuffer(outputData);return FAILED;}}INFO_LOG("create model output success");
/***************************************************/
/******************prepare input data***************/
/***************************************************/    if (modelDesc_ == nullptr) {ERROR_LOG("no model description, create input failed");return FAILED;}input_ = aclmdlCreateDataset();if (input_ == nullptr) {ERROR_LOG("can't create dataset, create input failed");return FAILED;}size_t modelInputSize = aclmdlGetInputSizeByIndex(modelDesc_, 0);ret = aclrtMalloc(&input_image_original, 784, ACL_MEM_MALLOC_NORMAL_ONLY);if (ret != ACL_SUCCESS) {ERROR_LOG("malloc device buffer failed. size is %zu, errorCode is %d",size, static_cast<int32_t>(ret));return FAILED;}unsigned char * input_image = static_cast<unsigned char*>(input_image_original);void* input_image_float_ori;ret = aclrtMalloc(&input_image_float_ori, 784*sizeof(float), ACL_MEM_MALLOC_NORMAL_ONLY);if (ret != ACL_SUCCESS) {ERROR_LOG("malloc device buffer failed. size is %zu, errorCode is %d",size, static_cast<int32_t>(ret));return FAILED;}float * input_image_float = static_cast<float*>(input_image_float_ori);;Load_data(num,input_image);for(int num=0;num<784;num++){input_image_float[num]=(float)input_image[num];}
/*     aclrtFree(input_image);input_image=nullptr; */aclDataBuffer *inputData = aclCreateDataBuffer(input_image_float, modelInputSize);if (inputData == nullptr) {ERROR_LOG("can't create data buffer, create input failed");return FAILED;}ret = aclmdlAddDatasetBuffer(input_, inputData);if (ret != ACL_SUCCESS) {ERROR_LOG("add input dataset buffer failed, errorCode is %d", static_cast<int32_t>(ret));(void)aclDestroyDataBuffer(inputData);inputData = nullptr;return FAILED;}INFO_LOG("create model input success");ret = aclrtMalloc(&time_ori, loop_count*sizeof(int64_t), ACL_MEM_MALLOC_NORMAL_ONLY);if (ret != ACL_SUCCESS) {ERROR_LOG("malloc device buffer failed. size is %zu, errorCode is %d",loop_count*sizeof(int64_t), static_cast<int32_t>(ret));return FAILED;}int64_t * time = static_cast<int64_t*>(time_ori);for(int loop_time=0;loop_time < loop_count;loop_time++){num++;Load_data(num,input_image);for(int loop_num=0;loop_num<784;loop_num++){input_image_float[loop_num]=(float)input_image[loop_num];}void* data = aclGetDataBufferAddr(inputData);uint32_t len = aclGetDataBufferSizeV2(inputData);     float *indata = NULL;  indata = reinterpret_cast<float*>(data);  /***************************************************//******************inference************************//***************************************************/start_time = getCurrentTimeUs();ret = aclmdlExecute(modelId_, input_, output_);end_time = getCurrentTimeUs();time[loop_time]=end_time-start_time;sum=sum+time[loop_time];printf("---Elapse Time = %.3f ms \n", (end_time-start_time) / 1000.f);   
/***************************************************/
/******************post process*********************/
/***************************************************/// get model output dataaclDataBuffer* dataBuffer = aclmdlGetDatasetBuffer(output_, 0);void* data_1 = aclGetDataBufferAddr(dataBuffer);uint32_t len_1 = aclGetDataBufferSizeV2(dataBuffer);     float *outData = NULL;  outData = reinterpret_cast<float*>(data_1);  void* buffer_copy_ori;ret = aclrtMalloc(&buffer_copy_ori, len_1*sizeof(float), ACL_MEM_MALLOC_NORMAL_ONLY);if (ret != ACL_SUCCESS) {ERROR_LOG("malloc device buffer failed. size is %zu, errorCode is %d",len_1, static_cast<int32_t>(ret));return FAILED;}float * buffer_copy = static_cast<float*>(buffer_copy_ori);for(int i_1 = 0; i_1 < len_1/sizeof(*outData);i_1++){buffer_copy[i_1]=outData[i_1];}Bubble_sort(outData,len_1/sizeof(*outData));for(int i_2 =0;i_2<len_1/sizeof(*outData);i_2++){if(buffer_copy[i_2]==outData[9]){printf("------------------------------------------%d time the pic value is %d \n",loop_time,i_2);}} aclrtFree(buffer_copy);buffer_copy=nullptr;}printf("--------loop %d times sum is %.4f ms average time is %.3f ms\n", loop_count,sum / 1000.f,(sum / 1000.f)/loop_count);aclrtFree(time);time=nullptr;aclrtFree(input_image);input_image=nullptr;aclrtFree(input_image_float);input_image_float=nullptr;/***************************************************/
/*******************destroy model input*************/
/***************************************************/for(size_t i = 0; i < aclmdlGetDatasetNumBuffers(input_); ++i) {aclDataBuffer *dataBuffer = aclmdlGetDatasetBuffer(input_, i);(void)aclDestroyDataBuffer(dataBuffer);}(void)aclmdlDestroyDataset(input_);input_ = nullptr;INFO_LOG("destroy model input success");
/***************************************************/
/*********************destroy model output*********/
/***************************************************/for (size_t i = 0; i < aclmdlGetDatasetNumBuffers(output_); ++i) {aclDataBuffer* dataBuffer = aclmdlGetDatasetBuffer(output_, i);void* data = aclGetDataBufferAddr(dataBuffer);(void)aclrtFree(data);(void)aclDestroyDataBuffer(dataBuffer);}(void)aclmdlDestroyDataset(output_);output_ = nullptr;INFO_LOG("destroy model output success");
/***************************************************/
/******uninstall model and release resource*********/
/***************************************************/modelId_=1;ret = aclmdlUnload(modelId_);// releasemodelDesc_if (modelDesc_ != nullptr) {aclmdlDestroyDesc(modelDesc_);modelDesc_ = nullptr;}INFO_LOG("unload model success, modelId is %u", modelId_);// release resorceif (stream_ != nullptr) {ret = aclrtDestroyStream(stream_);if (ret != ACL_SUCCESS) {ERROR_LOG("destroy stream failed, errorCode = %d", static_cast<int32_t>(ret));}stream_ = nullptr;}INFO_LOG("end to destroy stream");if (context_ != nullptr) {ret = aclrtDestroyContext(context_);if (ret != ACL_SUCCESS) {ERROR_LOG("destroy context failed, errorCode = %d", static_cast<int32_t>(ret));}context_ = nullptr;}INFO_LOG("end to destroy context");ret = aclrtResetDevice(deviceId_);if (ret != ACL_SUCCESS) {ERROR_LOG("reset device %d failed, errorCode = %d", deviceId_, static_cast<int32_t>(ret));}INFO_LOG("end to reset device %d", deviceId_);ret = aclFinalize();if (ret != ACL_SUCCESS) {ERROR_LOG("finalize acl failed, errorCode = %d", static_cast<int32_t>(ret));}INFO_LOG("end to finalize acl");
}

 

执行结果：

fp32

int8