Qiea/c-cnn
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

merge into: Qiea:main
Branches Tags
Qiea:main Qiea:mini
...
pull from: Qiea:mini
Branches Tags
Qiea:mini Qiea:main

3 Commits
main ... mini

Author SHA1 Message Date
Qiea
cbe00c6c95 删除 CMakeLists.txt 2024-12-18 11:30:57 +08:00
Qiea
c453296dd4 删除 .gitignore 2024-12-18 11:30:30 +08:00
Qiea
13d21f152c MINI版本 2024-12-18 11:28:00 +08:00
23 changed files with 5052734 additions and 359 deletions
Expand all files Collapse all files

14
.gitignore vendored
View File

@@ -1,14 +0,0 @@
CMakeLists.txt.user
CMakeCache.txt
CMakeFiles
CMakeScripts
Testing
Makefile
cmake_install.cmake
install_manifest.txt
compile_commands.json
CTestTestfile.cmake
_deps
.gitattributes
/cmake-build-debug
/.idea

13
CMakeLists.txt
View File

@@ -1,13 +0,0 @@
cmake_minimum_required(VERSION 3.26)
project(c_cnn C)
set(CMAKE_C_STANDARD 11)
include_directories(.)
add_executable(c_cnn
cnn.c
cnn.h
main.c
cnn_model.c
debug.c)

2
README.md
View File

@@ -1,2 +0,0 @@
# c-cnn
c-cnn

BIN
c_cnn.exe Normal file
View File

Binary file not shown.

188
cnn.c
View File

@@ -2,7 +2,7 @@
void print_rslt(float* rslt, u8 input_matrix_length, u32 length){ void print_rslt(float* rslt, uint8_t input_matrix_length, uint32_t length){
int _tmp = 0; int _tmp = 0;
printf("[0:0]"); printf("[0:0]");
for (int i = 0; i < length; i++) { for (int i = 0; i < length; i++) {
@@ -14,7 +14,7 @@ void print_rslt(float* rslt, u8 input_matrix_length, u32 length){
printf("\r\n"); printf("\r\n");
} }
// 将原始矩阵复制到填充后的矩阵中央 // <EFBFBD><EFBFBD>ԭʼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ƶ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ľ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
float* expand(const float* old_matrix, int old_matrix_length, int layer){ float* expand(const float* old_matrix, int old_matrix_length, int layer){
float* new_matrix = (float *)malloc(sizeof(float)*layer*(old_matrix_length+2)*(old_matrix_length+2)); float* new_matrix = (float *)malloc(sizeof(float)*layer*(old_matrix_length+2)*(old_matrix_length+2));
memset(new_matrix, 0, sizeof(float)*layer*(old_matrix_length+2)*(old_matrix_length+2)); memset(new_matrix, 0, sizeof(float)*layer*(old_matrix_length+2)*(old_matrix_length+2));
@@ -31,34 +31,34 @@ float* expand(const float* old_matrix, int old_matrix_length, int layer){
return new_matrix; return new_matrix;
} }
//model 模型名字 //model ģ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
//input_matrix 输入图像 //input_matrix <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD>
//input_matrix_length 输入图像的边长：102 //input_matrix_length <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD>ı߳<EFBFBD><EFBFBD><EFBFBD>102
//c_rl 输出图像的边长：100 //c_rl <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD>ı߳<EFBFBD><EFBFBD><EFBFBD>100
//返回卷积的结果 //<EFBFBD><EFBFBD><EFBFBD>ؾ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ľ<EFBFBD><EFBFBD><EFBFBD>
float* convolution(Model model_w, Model model_b, const float* input_matrix, int input_matrix_length){ float* convolution(Model model_w, Model model_b, const float* input_matrix, int input_matrix_length){
// 初始化卷积层参数 // <EFBFBD><EFBFBD>ʼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
int im_l = input_matrix_length; int im_l = input_matrix_length;
int cr_l = input_matrix_length - 2; int cr_l = input_matrix_length - 2;
float conv_temp; // 临时变量，用于存储卷积计算的中间结果 float conv_temp; // <EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>м<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
//用于合并前的数组，具有32*64*50*50(第二层)的大小 //<EFBFBD><EFBFBD><EFBFBD>ںϲ<EFBFBD>ǰ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>飬<EFBFBD><EFBFBD><EFBFBD><EFBFBD>32*64*50*50(<EFBFBD>ڶ<EFBFBD><EFBFBD><EFBFBD>)<29>Ĵ<EFBFBD>С
float* _conv_rlst = (float *) malloc(sizeof (float) * model_w.num_kernels * (cr_l * cr_l)); float* _conv_rlst = (float *) malloc(sizeof (float) * model_w.num_kernels * (cr_l * cr_l));
memset(_conv_rlst, 0, sizeof (float) * model_w.num_kernels * (cr_l * cr_l)); memset(_conv_rlst, 0, sizeof (float) * model_w.num_kernels * (cr_l * cr_l));
//子图合并后的数组 //<EFBFBD><EFBFBD>ͼ<EFBFBD>ϲ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
float* conv_rlst = (float *) malloc(sizeof (float) * model_w.num_kernels * (cr_l * cr_l)); float* conv_rlst = (float *) malloc(sizeof (float) * model_w.num_kernels * (cr_l * cr_l));
memset(conv_rlst, 0, sizeof (float) * model_w.num_kernels * (cr_l * cr_l)); memset(conv_rlst, 0, sizeof (float) * model_w.num_kernels * (cr_l * cr_l));
// 遍历30个卷积核（假设有30个通道） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD>30<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ˣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>30<EFBFBD><EFBFBD>ͨ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>
for(int c=0; c < model_w.channel; c++){ for(int c=0; c < model_w.channel; c++){
for(int k=0; k < model_w.num_kernels; k++){ for(int k=0; k < model_w.num_kernels; k++){
for(int row = 0; row < cr_l; row++) { for(int row = 0; row < cr_l; row++) {
for (int col = 0; col < cr_l; col++) { for (int col = 0; col < cr_l; col++) {
conv_temp = 0; // 每个输出像素初始化为0 conv_temp = 0; // ÿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>س<EFBFBD>ʼ<EFBFBD><EFBFBD>Ϊ0
// 进行3x3的卷积操作 // <EFBFBD><EFBFBD><EFBFBD><EFBFBD>3x3<EFBFBD>ľ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
for (int x = 0; x < 3; x++) { for (int x = 0; x < 3; x++) {
for (int y = 0; y < 3; y++) { for (int y = 0; y < 3; y++) {
// 将输入图像的对应像素与卷积核权重相乘，并累加到conv_temp // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD>Ķ<EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ȩ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ˣ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ۼӵ<EFBFBD>conv_temp
conv_temp += input_matrix[(c*im_l*im_l) + (row*(im_l)+col) + (x*(im_l)+y)] conv_temp += input_matrix[(c*im_l*im_l) + (row*(im_l)+col) + (x*(im_l)+y)]
* model_w.array[((c+k*model_w.channel)*3*3) + (x*3+y)]; * model_w.array[((c+k*model_w.channel)*3*3) + (x*3+y)];
} }
@@ -67,7 +67,7 @@ float* convolution(Model model_w, Model model_b, const float* input_matrix, int
} }
} }
} }
//合并子图 //<EFBFBD>ϲ<EFBFBD><EFBFBD><EFBFBD>ͼ
for(int k=0; k < model_w.num_kernels; k++) { for(int k=0; k < model_w.num_kernels; k++) {
for (int row = 0; row < cr_l; row++) { for (int row = 0; row < cr_l; row++) {
for (int col = 0; col < cr_l; col++) { for (int col = 0; col < cr_l; col++) {
@@ -80,13 +80,13 @@ float* convolution(Model model_w, Model model_b, const float* input_matrix, int
for (int row = 0; row < cr_l; row++) { for (int row = 0; row < cr_l; row++) {
for (int col = 0; col < cr_l; col++) { for (int col = 0; col < cr_l; col++) {
conv_temp = conv_rlst[(k * cr_l * cr_l) + (row * cr_l + col)]; conv_temp = conv_rlst[(k * cr_l * cr_l) + (row * cr_l + col)];
// 加上对应卷积核的偏置 // <EFBFBD><EFBFBD><EFBFBD>϶<EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˵<EFBFBD>ƫ<EFBFBD><EFBFBD>
conv_temp += model_b.array[k]; conv_temp += model_b.array[k];
// 激活函数：ReLU（将小于0的值设为0） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ReLU<EFBFBD><EFBFBD><EFBFBD><EFBFBD>С<EFBFBD><EFBFBD>0<EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD>Ϊ0<EFBFBD><EFBFBD>
if (conv_temp > 0) if (conv_temp > 0)
conv_rlst[(k * (cr_l * cr_l)) + (row * cr_l) + (col)] = conv_temp; // 如果卷积结果大于0，存入结果数组 conv_rlst[(k * (cr_l * cr_l)) + (row * cr_l) + (col)] = conv_temp; // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>0<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
else else
conv_rlst[(k * (cr_l * cr_l)) + (row * cr_l) + (col)] = 0; // 否则存入0 conv_rlst[(k * (cr_l * cr_l)) + (row * cr_l) + (col)] = 0; // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>0
} }
} }
} }
@@ -95,38 +95,38 @@ float* convolution(Model model_w, Model model_b, const float* input_matrix, int
return conv_rlst; return conv_rlst;
} }
//num_kernels 卷积核的个数：32 //num_kernels <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>˵ĸ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>32
//area 池化的面积：2*2 //area <EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>2*2
//input_matrix 输入图像 //input_matrix <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD>
//input_matrix_length 输入图像的边长：100 //input_matrix_length <EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD>ı߳<EFBFBD><EFBFBD><EFBFBD>100
//输出图像的边长：50 //<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD>ı߳<EFBFBD><EFBFBD><EFBFBD>50
//返回池化的结果 //<EFBFBD><EFBFBD><EFBFBD>سػ<EFBFBD><EFBFBD>Ľ<EFBFBD><EFBFBD><EFBFBD>
float* pooling(Model model_w, const float* input_matrix, u8 input_matrix_length){ float* pooling(Model model_w, const float* input_matrix, uint8_t input_matrix_length){
u8 im_l = input_matrix_length; uint8_t im_l = input_matrix_length;
float pool_temp = 0; // 临时变量，用于存储池化操作的最大值 float pool_temp = 0; // <EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ洢<EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ
float* pool_rslt = (float *) malloc(sizeof (float)*model_w.num_kernels*im_l*im_l); float* pool_rslt = (float *) malloc(sizeof (float)*model_w.num_kernels*im_l*im_l);
memset(pool_rslt, 0, sizeof (float)*model_w.num_kernels*im_l*im_l); memset(pool_rslt, 0, sizeof (float)*model_w.num_kernels*im_l*im_l);
// 遍历30个通道（与卷积核数量相同） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD>30<EFBFBD><EFBFBD>ͨ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͬ<EFBFBD><EFBFBD>
for(u8 n=0; n<model_w.num_kernels; n++) for(uint8_t n=0; n<model_w.num_kernels; n++)
{ {
// 遍历输入图像的每一行，步长为2（2x2的池化窗口） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ÿһ<EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ2<EFBFBD><EFBFBD>2x2<EFBFBD>ijػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڣ<EFBFBD>
for(u8 row=0; row<im_l; row=row+2) for(uint8_t row=0; row<im_l; row=row+2)
{ {
// 遍历输入图像的每一列，步长为2 // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ͼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ÿһ<EFBFBD>У<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ2
for(u8 col=0; col<im_l; col=col+2) for(uint8_t col=0; col<im_l; col=col+2)
{ {
pool_temp = 0; // 每个池化区域的最大值初始化为0 pool_temp = 0; // ÿ<EFBFBD><EFBFBD><EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD>ʼ<EFBFBD><EFBFBD>Ϊ0
// 进行2x2的最大池化操作 // <EFBFBD><EFBFBD><EFBFBD><EFBFBD>2x2<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
for(u8 x=0; x<2; x++) for(uint8_t x=0; x<2; x++)
{ {
for(u8 y=0; y<2; y++) for(uint8_t y=0; y<2; y++)
{ {
// 更新当前池化区域的最大值 // <EFBFBD><EFBFBD><EFBFBD>µ<EFBFBD>ǰ<EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ
if(pool_temp <= input_matrix[row*im_l+col+x*im_l+y+n*(im_l*im_l)]) if(pool_temp <= input_matrix[row*im_l+col+x*im_l+y+n*(im_l*im_l)])
pool_temp = input_matrix[row*im_l+col+x*im_l+y+n*(im_l*im_l)]; pool_temp = input_matrix[row*im_l+col+x*im_l+y+n*(im_l*im_l)];
} }
} }
// 将最大值存入池化结果数组 // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
pool_rslt[(row/2)*(im_l/2)+col/2+n*((im_l/2)*(im_l/2))] = pool_temp; pool_rslt[(row/2)*(im_l/2)+col/2+n*((im_l/2)*(im_l/2))] = pool_temp;
} }
} }
@@ -135,29 +135,29 @@ float* pooling(Model model_w, const float* input_matrix, u8 input_matrix_length)
} }
float* hidden(const float* input_matrix){ float* hidden(const float* input_matrix){
float affine1_temp; // 临时变量，用于存储全连接层的中间结果 float affine1_temp; // <EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ洢ȫ<EFBFBD><EFBFBD><EFBFBD>Ӳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>м<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
float *affine1_rslt = (float *) malloc(sizeof(float)*128); float *affine1_rslt = (float *) malloc(sizeof(float)*128);
memset(affine1_rslt, 0, sizeof(float)*128); memset(affine1_rslt, 0, sizeof(float)*128);
// 遍历128个神经元（假设隐藏层有128个神经元） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD>128<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ز<EFBFBD><EFBFBD><EFBFBD>128<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD>
for(u8 n=0; n<128; n++) for(uint8_t n=0; n<128; n++)
{ {
affine1_temp = 0; // 每个神经元的输出初始化为0 affine1_temp = 0; // ÿ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʼ<EFBFBD><EFBFBD>Ϊ0
// 进行矩阵乘法，将池化层输出展平为一维向量后，与全连接层权重进行点积 // <EFBFBD><EFBFBD><EFBFBD>о<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ػ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>չƽΪһά<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ȫ<EFBFBD><EFBFBD><EFBFBD>Ӳ<EFBFBD>Ȩ<EFBFBD>ؽ<EFBFBD><EFBFBD>е<EFBFBD><EFBFBD><EFBFBD>
for(int i=0; i<(128*12*12); i++) for(int i=0; i<(128*12*12); i++)
{ {
affine1_temp = affine1_temp + input_matrix[i] * fc1_weight.array[i+(128*12*12)*n]; affine1_temp = affine1_temp + input_matrix[i] * fc1_weight.array[i+(128*12*12)*n];
} }
// 加上对应神经元的偏置 // <EFBFBD><EFBFBD><EFBFBD>϶<EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD>ƫ<EFBFBD><EFBFBD>
affine1_temp = affine1_temp + fc1_bias.array[n]; affine1_temp = affine1_temp + fc1_bias.array[n];
// 激活函数：ReLU（将小于0的值设为0） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ReLU<EFBFBD><EFBFBD><EFBFBD><EFBFBD>С<EFBFBD><EFBFBD>0<EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD>Ϊ0<EFBFBD><EFBFBD>
if(affine1_temp > 0) if(affine1_temp > 0)
affine1_rslt[n] = affine1_temp; // 如果结果大于0，存入结果数组 affine1_rslt[n] = affine1_temp; // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>0<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
else else
affine1_rslt[n] = 0; // 否则存入0 affine1_rslt[n] = 0; // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>0
} }
// print_rslt(affine1_rslt,1,128); // print_rslt(affine1_rslt,1,128);
@@ -166,25 +166,25 @@ float* hidden(const float* input_matrix){
} }
float* output(Model model_w, const float* input_matrix){ float* output(Model model_w, const float* input_matrix){
u8 num = model_w.num_kernels; uint8_t num = model_w.num_kernels;
float affine2_temp; // 临时变量，用于存储输出层的中间结果 float affine2_temp; // <EFBFBD><EFBFBD>ʱ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ڴ洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>м<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
float *affine2_rslt = (float *) malloc(sizeof(float)*num); float *affine2_rslt = (float *) malloc(sizeof(float)*num);
memset(affine2_rslt, 0, sizeof(float)*num); memset(affine2_rslt, 0, sizeof(float)*num);
// 遍历10个输出神经元（假设有10个类别） // <EFBFBD><EFBFBD><EFBFBD><EFBFBD>10<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>10<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
for(int n=0; n<num; n++) for(int n=0; n<num; n++)
{ {
affine2_temp = 0; // 当前神经元的输出初始化为0 affine2_temp = 0; // <EFBFBD><EFBFBD>ǰ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ʼ<EFBFBD><EFBFBD>Ϊ0
// 进行矩阵乘法，将隐藏层的输出与输出层权重进行点积 // <EFBFBD><EFBFBD><EFBFBD>о<EFBFBD><EFBFBD><EFBFBD><EFBFBD>˷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ز<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ȩ<EFBFBD>ؽ<EFBFBD><EFBFBD>е<EFBFBD><EFBFBD><EFBFBD>
for(int i=0; i<128; i++) for(int i=0; i<128; i++)
{ {
affine2_temp = affine2_temp + fc2_weight.array[i+128*n] * input_matrix[i]; affine2_temp = affine2_temp + fc2_weight.array[i+128*n] * input_matrix[i];
} }
// 加上对应神经元的偏置 // <EFBFBD><EFBFBD><EFBFBD>϶<EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD>ƫ<EFBFBD><EFBFBD>
affine2_temp = affine2_temp + fc2_weight.array[n]; affine2_temp = affine2_temp + fc2_weight.array[n];
affine2_rslt[n] = affine2_temp; // 存储输出层的结果 affine2_rslt[n] = affine2_temp; // <EFBFBD>洢<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ľ<EFBFBD><EFBFBD><EFBFBD>
} }
return affine2_rslt; return affine2_rslt;
@@ -214,7 +214,7 @@ void calculate_statistics(Model model, float* value)
value[2] = sqrtf(variance); value[2] = sqrtf(variance);
} }
u8 check_threshold(Model model, const float* value) uint8_t check_threshold(Model model, const float* value)
{ {
const float threshold = 20; const float threshold = 20;
@@ -232,7 +232,7 @@ float* generateMatrix(Model model, const float* value)
float* CNN_data = (float*) malloc(sizeof(float)*100*100); float* CNN_data = (float*) malloc(sizeof(float)*100*100);
memset(CNN_data, 0, sizeof(float)*100*100); memset(CNN_data, 0, sizeof(float)*100*100);
u16 x = model.maxlength / 100; uint16_t x = model.maxlength / 100;
float y = value[0] / 100; float y = value[0] / 100;
for (int i = 0; i < model.maxlength; i++) { for (int i = 0; i < model.maxlength; i++) {
@@ -256,7 +256,7 @@ float* generateMatrix(Model model, const float* value)
float calculate_probabilities(Model model_w, float *input_array) float calculate_probabilities(Model model_w, float *input_array)
{ {
float sum = 0; float sum = 0;
u8 input_num = model_w.num_kernels; uint8_t input_num = model_w.num_kernels;
float *result = (float *) malloc(sizeof(float)*input_num); float *result = (float *) malloc(sizeof(float)*input_num);
memset(result, 0, sizeof(float)*input_num); memset(result, 0, sizeof(float)*input_num);
@@ -294,16 +294,16 @@ float calculate_probabilities(Model model_w, float *input_array)
} }
u8 calculate_layer(Model model_w, float *input_array){ uint8_t calculate_layer(Model model_w, float *input_array){
u8 input_num = model_w.num_kernels; uint8_t input_num = model_w.num_kernels;
u8 predict_num = 0; uint8_t predict_num = 0;
float max_temp = -100; float max_temp = -100;
for(int n=0; n<input_num; n++) for(int n=0; n<input_num; n++)
{ {
if(max_temp <= input_array[n]) if(max_temp <= input_array[n])
{ {
max_temp = input_array[n]; // 更新最大值 max_temp = input_array[n]; // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ
predict_num = n; // 记录最大值对应的类别索引 predict_num = n; // <EFBFBD><EFBFBD>¼<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ<EFBFBD><EFBFBD>Ӧ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
} }
} }
print_rslt(input_array,input_num,input_num); print_rslt(input_array,input_num,input_num);
@@ -312,13 +312,12 @@ u8 calculate_layer(Model model_w, float *input_array){
void cnn_run(){ void cnn_run(){
#if is1300000
float value[3] = {0}; float value[3] = {0};
calculate_statistics(data,&value[0]); calculate_statistics(data,&value[0]);
if (check_threshold(data,&value[0])){ if (check_threshold(data,&value[0])){
float* _data = generateMatrix(data,&value[0]); float* _data = generateMatrix(data,&value[0]);
//第一层：填充102 * 102 //<EFBFBD><EFBFBD>һ<EFBFBD>㣺<EFBFBD><EFBFBD><EFBFBD><EFBFBD>102 * 102
float* expand_matrix_1 = expand(_data, 100, 1); float* expand_matrix_1 = expand(_data, 100, 1);
float* conv_rlst_1 = convolution(conv1_weight,conv1_bias,expand_matrix_1, 102); float* conv_rlst_1 = convolution(conv1_weight,conv1_bias,expand_matrix_1, 102);
float* pool_rslt_1 = pooling(conv1_weight, conv_rlst_1, 100); float* pool_rslt_1 = pooling(conv1_weight, conv_rlst_1, 100);
@@ -330,7 +329,7 @@ void cnn_run(){
free(conv_rlst_1); free(conv_rlst_1);
conv_rlst_1 = NULL; conv_rlst_1 = NULL;
//第二层：填充32 * 52 * 52 //<EFBFBD>ڶ<EFBFBD><EFBFBD>㣺<EFBFBD><EFBFBD><EFBFBD><EFBFBD>32 * 52 * 52
float* expand_matrix_2 = expand(pool_rslt_1, 50, 32); float* expand_matrix_2 = expand(pool_rslt_1, 50, 32);
float* conv_rlst_2 = convolution(conv2_weight,conv2_bias,expand_matrix_2, 52); float* conv_rlst_2 = convolution(conv2_weight,conv2_bias,expand_matrix_2, 52);
float* pool_rslt_2 = pooling(conv2_weight, conv_rlst_2, 50); float* pool_rslt_2 = pooling(conv2_weight, conv_rlst_2, 50);
@@ -342,7 +341,7 @@ void cnn_run(){
free(conv_rlst_2); free(conv_rlst_2);
conv_rlst_2 = NULL; conv_rlst_2 = NULL;
//第三层：填充 64 * 27 * 27 //<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>㣺<EFBFBD><EFBFBD><EFBFBD><EFBFBD> 64 * 27 * 27
float* expand_matrix_3 = expand(pool_rslt_2, 25, 64); float* expand_matrix_3 = expand(pool_rslt_2, 25, 64);
float* conv_rlst_3 = convolution(conv3_weight,conv3_bias,expand_matrix_3, 27); float* conv_rlst_3 = convolution(conv3_weight,conv3_bias,expand_matrix_3, 27);
float* pool_rslt_3 = pooling(conv3_weight, conv_rlst_3, 25); float* pool_rslt_3 = pooling(conv3_weight, conv_rlst_3, 25);
@@ -357,7 +356,7 @@ void cnn_run(){
float* affine1_rslt = hidden(pool_rslt_3); float* affine1_rslt = hidden(pool_rslt_3);
float* affine2_rslt = output(fc2_weight, affine1_rslt); float* affine2_rslt = output(fc2_weight, affine1_rslt);
printf("概率：%f\r\n",calculate_probabilities(fc2_weight, affine2_rslt)); printf("<EFBFBD><EFBFBD><EFBFBD>ʣ<EFBFBD>%f\r\n",calculate_probabilities(fc2_weight, affine2_rslt));
printf("Label is:%d\r\n",calculate_layer(fc2_weight, affine2_rslt)); printf("Label is:%d\r\n",calculate_layer(fc2_weight, affine2_rslt));
free(pool_rslt_3); free(pool_rslt_3);
@@ -368,53 +367,6 @@ void cnn_run(){
affine2_rslt = NULL; affine2_rslt = NULL;
} else{ } else{
printf("未放电！最大值:%f 平均值:%f 标准差:%f\r\n",value[0],value[1],value[2]); printf("δ<EFBFBD>ŵ磡<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ֵ:%f ƽ<><C6BD>ֵ:%f <20><>׼<EFBFBD><D7BC>:%f\r\n",value[0],value[1],value[2]);
} }
#else
float* expand_matrix_1 = expand(data.array, 100, 1);
float* conv_rlst_1 = convolution(conv1_weight,conv1_bias,expand_matrix_1, 102);
float* pool_rslt_1 = pooling(conv1_weight, conv_rlst_1, 100);
free(expand_matrix_1);
expand_matrix_1 = NULL;
free(conv_rlst_1);
conv_rlst_1 = NULL;
//第二层：填充32 * 52 * 52
float* expand_matrix_2 = expand(pool_rslt_1, 50, 32);
float* conv_rlst_2 = convolution(conv2_weight,conv2_bias,expand_matrix_2, 52);
float* pool_rslt_2 = pooling(conv2_weight, conv_rlst_2, 50);
free(pool_rslt_1);
pool_rslt_1 = NULL;
free(expand_matrix_2);
expand_matrix_2 = NULL;
free(conv_rlst_2);
conv_rlst_2 = NULL;
//第三层：填充 64 * 27 * 27
float* expand_matrix_3 = expand(pool_rslt_2, 25, 64);
float* conv_rlst_3 = convolution(conv3_weight,conv3_bias,expand_matrix_3, 27);
float* pool_rslt_3 = pooling(conv3_weight, conv_rlst_3, 25);
free(pool_rslt_2);
pool_rslt_2 = NULL;
free(expand_matrix_3);
expand_matrix_3 = NULL;
free(conv_rlst_3);
conv_rlst_3 = NULL;
float* affine1_rslt = hidden(pool_rslt_3);
float* affine2_rslt = output(fc2_weight, affine1_rslt);
printf("概率：%f\r\n",calculate_probabilities(fc2_weight, affine2_rslt));
printf("Label is:%d\r\n",calculate_layer(fc2_weight, affine2_rslt));
free(pool_rslt_3);
pool_rslt_3 = NULL;
free(affine1_rslt);
affine1_rslt = NULL;
free(affine2_rslt);
affine2_rslt = NULL;
#endif
} }

7
cnn.h
View File

@@ -1,14 +1,7 @@
#ifndef _CNN_H_ #ifndef _CNN_H_
#define _CNN_H_ #define _CNN_H_
#include "cnn_model.h" #include "cnn_model.h"
#include <stdio.h>
#include <math.h>
void cnn_run(void); void cnn_run(void);
#endif #endif

77
cnn_model.c
View File

@@ -57,7 +57,7 @@ float* modelmym_init(char* model_name){
u8 modelmym_free(char* model_name){ uint8_t modelmym_free(char* model_name){
if(conv1_bias.array != NULL && strcmp(model_name, "conv1_bias") == 0){ if(conv1_bias.array != NULL && strcmp(model_name, "conv1_bias") == 0){
free(conv1_bias.array); free(conv1_bias.array);
conv1_bias.array = NULL; conv1_bias.array = NULL;
@@ -143,7 +143,7 @@ u8 modelmym_free(char* model_name){
u8 model_write(char* model_name) uint8_t model_write(char* model_name)
{ {
if(strcmp(model_name, "all") == 0){ if(strcmp(model_name, "all") == 0){
model_write("conv1_bias"); model_write("conv1_bias");
@@ -158,11 +158,10 @@ u8 model_write(char* model_name)
model_write("fc2_weight"); model_write("fc2_weight");
model_info("all"); model_info("all");
SDRAM_USED();
}else{ }else{
u8 _times=0; uint8_t _times=0;
u32 _larr = 0; uint32_t _larr = 0;
u8 _len = strlen(model_name); uint8_t _len = strlen(model_name);
char _path[_len+1+7+30]; char _path[_len+1+7+30];
char _datapath[_len+1+7+30]; char _datapath[_len+1+7+30];
char _fstr[READLENGTH+1] = {0}; char _fstr[READLENGTH+1] = {0};
@@ -173,7 +172,7 @@ u8 model_write(char* model_name)
sprintf(_path, "./dataset/_data/%s.txt", model_name); sprintf(_path, "./dataset/_data/%s.txt", model_name);
FILE *file = fopen(_path, "r"); FILE *file = fopen(_path, "r");
if(file == NULL){ if(file == NULL){
DEBUG_PRINTF("\r\n输入了一个无效的模型或Data数据集的名字\r\n"); printf("\r\n<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>һ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ч<EFBFBD><EFBFBD>ģ<EFBFBD>ͻ<EFBFBD>Data<EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\r\n");
return 199; return 199;
}else{ }else{
_model = model("data"); _model = model("data");
@@ -185,13 +184,13 @@ u8 model_write(char* model_name)
sprintf(_path, "./dataset/%s.txt", _model -> name); sprintf(_path, "./dataset/%s.txt", _model -> name);
FILE *file = fopen(_path, "r"); FILE *file = fopen(_path, "r");
if(file == NULL){ if(file == NULL){
DEBUG_PRINTF("预设里没有这个模型：[%s]\r\n", _path); printf("Ԥ<EFBFBD><EFBFBD><EFBFBD><EFBFBD>û<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD>ͣ<EFBFBD>[%s]\r\n", _path);
return 4; return 4;
} }
} }
if(_model -> array == NULL && modelmym_init(_model -> name) == NULL){ if(_model -> array == NULL && modelmym_init(_model -> name) == NULL){
DEBUG_PRINTF("无法创建模型参数[%s]的数组到SDRAM里\r\n", _model -> name); printf("<EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD>Ͳ<EFBFBD><EFBFBD><EFBFBD>[%s]<5D><><EFBFBD><EFBFBD><EFBFBD>鵽SDRAM<EFBFBD><EFBFBD>\r\n", _model -> name);
return 200; return 200;
} }
@@ -199,40 +198,40 @@ u8 model_write(char* model_name)
sprintf(_path, "./dataset/%s.txt", _model -> dname ? _datapath : _model -> name); sprintf(_path, "./dataset/%s.txt", _model -> dname ? _datapath : _model -> name);
FILE *file = fopen(_path, "r"); FILE *file = fopen(_path, "r");
if(file == NULL){ if(file == NULL){
DEBUG_PRINTF("文件[%s]无法打开\r\n", _model -> dname ? _model -> dname : _model -> name); printf("<EFBFBD>ļ<EFBFBD>[%s]<EFBFBD>޷<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\r\n", _model -> dname ? _model -> dname : _model -> name);
return 199; return 199;
} }
DEBUG_PRINTF("写入的模型参数名字是：%s\r\n", _model -> name); printf("д<EFBFBD><EFBFBD><EFBFBD><EFBFBD>ģ<EFBFBD>Ͳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ǣ<EFBFBD>%s\r\n", _model -> name);
if(_model -> dname)DEBUG_PRINTF("写入的Data数据集是：%s\r\n", _model -> dname); if(_model -> dname)printf("д<EFBFBD><EFBFBD><EFBFBD><EFBFBD>Data<EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD><EFBFBD>ǣ<EFBFBD>%s\r\n", _model -> dname);
DEBUG_PRINTF("写入模型参数数组的最大长度为：%d\r\n", _model -> maxlength); printf("д<EFBFBD><EFBFBD>ģ<EFBFBD>Ͳ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>󳤶<EFBFBD>Ϊ<EFBFBD><EFBFBD>%d\r\n", _model -> maxlength);
DEBUG_PRINTF("目前数组存活的元素数量为：%d", _model -> realength); printf("Ŀǰ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ԫ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD>%d", _model -> realength);
printf("\r\n正在写入模型参数'%s'，请稍后......\r\n",_model -> dname ? _model -> dname : _model -> name); printf("\r\n<EFBFBD><EFBFBD><EFBFBD><EFBFBD>д<EFBFBD><EFBFBD>ģ<EFBFBD>Ͳ<EFBFBD><EFBFBD><EFBFBD>'%s'<27><><EFBFBD><EFBFBD><EFBFBD>Ժ<EFBFBD>......\r\n",_model -> dname ? _model -> dname : _model -> name);
while(_larr < _model->maxlength && fgets(_fstr, sizeof(_fstr), file) != NULL){ while(_larr < _model->maxlength && fgets(_fstr, sizeof(_fstr), file) != NULL){
char *endptr; char *endptr;
float value = strtof(_fstr, &endptr); float value = strtof(_fstr, &endptr);
// 检查是否转换成功 // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ƿ<EFBFBD>ת<EFBFBD><EFBFBD><EFBFBD>ɹ<EFBFBD>
if (endptr == _fstr) { if (endptr == _fstr) {
fprintf(stderr, "第 %d 行不是有效的浮点数: %s", _larr + 1, _fstr); fprintf(stderr, "<EFBFBD><EFBFBD> %d <EFBFBD>в<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ч<EFBFBD>ĸ<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>: %s", _larr + 1, _fstr);
continue; // 跳过无效行 continue; // <EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>Ч<EFBFBD><EFBFBD>
} }
_model->array[_larr++] = value; _model->array[_larr++] = value;
_model->realength++; _model->realength++;
if(_model -> maxlength >= 73728 && (_larr >= (_model -> maxlength/10)*_times)){ if(_model -> maxlength >= 73728 && (_larr >= (_model -> maxlength/10)*_times)){
progress = _larr >= _model -> maxlength ? 100 : _times++ == 0 ? 0 : progress + 10; progress = _larr >= _model -> maxlength ? 100 : _times++ == 0 ? 0 : progress + 10;
DEBUG_PRINTF("\r\n["); printf("\r\n[");
for(u16 j=0; j<50;j++){ for(uint16_t j=0; j<50;j++){
if(j < progress/2) DEBUG_PRINTF("="); if(j < progress/2) printf("=");
else DEBUG_PRINTF(" "); else printf(" ");
} }
DEBUG_PRINTF("] %d%%", progress); printf("] %d%%", progress);
} }
if(_larr >= _model -> maxlength)break; if(_larr >= _model -> maxlength)break;
} }
DEBUG_PRINTF("\r\n模型参数[%s]已写入到内存中! 模型长度为 %d\r\n",_model -> dname ? _model -> dname : _model -> name,_model -> realength); printf("\r\nģ<EFBFBD>Ͳ<EFBFBD><EFBFBD><EFBFBD>[%s]<5D><>д<EFBFBD>뵽<EFBFBD>ڴ<EFBFBD><DAB4><EFBFBD>! ģ<>ͳ<EFBFBD><CDB3><EFBFBD>Ϊ %d\r\n",_model -> dname ? _model -> dname : _model -> name,_model -> realength);
return 0; return 0;
} }
return 1; return 1;
@@ -240,7 +239,7 @@ u8 model_write(char* model_name)
u8 model_read(char* model_name, u32 start, u32 end, u32 gap){ uint8_t model_read(char* model_name, uint32_t start, uint32_t end, uint32_t gap){
if(strcmp(model_name, "all") == 0){ if(strcmp(model_name, "all") == 0){
model_read("conv1_bias", start, end, gap); model_read("conv1_bias", start, end, gap);
model_read("conv1_weight", start, end, gap); model_read("conv1_weight", start, end, gap);
@@ -257,7 +256,7 @@ u8 model_read(char* model_name, u32 start, u32 end, u32 gap){
Model *_model = model(model_name); Model *_model = model(model_name);
if(_model == NULL || end == 0 || start > _model -> realength || end > _model -> realength)return 0; if(_model == NULL || end == 0 || start > _model -> realength || end > _model -> realength)return 0;
if(_model -> realength){ if(_model -> realength){
for (u32 i=0;i<((end > (_model -> realength) ? _model -> realength : (end-start))+(end > (_model -> realength) ? _model -> realength : (end-start)%gap ? 2 : 1));i+=gap) for (uint32_t i=0;i<((end > (_model -> realength) ? _model -> realength : (end-start))+(end > (_model -> realength) ? _model -> realength : (end-start)%gap ? 2 : 1));i+=gap)
printf("\r\n%s_floatArray[%d]: %f",_model->name,(i+start)<_model->realength ? i+start : _model->realength-1,(i+start)<_model->realength ? _model->array[i+start] : _model->array[_model -> realength-1]); printf("\r\n%s_floatArray[%d]: %f",_model->name,(i+start)<_model->realength ? i+start : _model->realength-1,(i+start)<_model->realength ? _model->array[i+start] : _model->array[_model -> realength-1]);
printf("\r\n"); printf("\r\n");
return 1; return 1;
@@ -268,30 +267,30 @@ u8 model_read(char* model_name, u32 start, u32 end, u32 gap){
u8 model_switchdata(char* data_name){ uint8_t model_switchdata(char* data_name){
u8 _len = strlen(data_name); uint8_t _len = strlen(data_name);
char _path[_len+1+7]; char _path[_len+1+7];
if(data.array != NULL)modelmym_free("data"); if(data.array != NULL)modelmym_free("data");
sprintf(_path, "./dataset/_data/%s.txt", data_name); sprintf(_path, "./dataset/_data/%s.txt", data_name);
FILE *file = fopen(_path, "r"); FILE *file = fopen(_path, "r");
if(file == NULL){ if(file == NULL){
DEBUG_PRINTF("\r\nData数据集[%s]不存在\r\n",data_name); printf("\r\nData<EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD>[%s]<5D><><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\r\n",data_name);
return 0; return 0;
}else{ }else{
u8 _res = model_write(data_name); uint8_t _res = model_write(data_name);
if (_res) { if (_res) {
DEBUG_PRINTF("Data数据集[%s]切换失败！！\r\n",data_name); printf("Data<EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD>[%s]<5D>л<EFBFBD>ʧ<EFBFBD>ܣ<EFBFBD><DCA3><EFBFBD>\r\n",data_name);
return 0; return 0;
} }
else DEBUG_PRINTF("Data数据集[%s]切换成功！\r\n",data_name); else printf("Data<EFBFBD><EFBFBD><EFBFBD>ݼ<EFBFBD>[%s]<5D>л<EFBFBD><D0BB>ɹ<EFBFBD><C9B9><EFBFBD>\r\n",data_name);
DEBUG_PRINTF("data_name的长度为：%d\r\n_path的长度为：%d\r\n_path为：%s\r\n",_len,sizeof(_path),_path); printf("data_name<EFBFBD>ij<EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD>%d\r\n_path<EFBFBD>ij<EFBFBD><EFBFBD><EFBFBD>Ϊ<EFBFBD><EFBFBD>%d\r\n_pathΪ<EFBFBD><EFBFBD>%s\r\n",_len,sizeof(_path),_path);
return 1; return 1;
} }
} }
u8 model_info(char* model_name){ uint8_t model_info(char* model_name){
if(strcmp(model_name, "all") == 0){ if(strcmp(model_name, "all") == 0){
model_info("conv1_bias"); model_info("conv1_bias");
model_info("conv1_weight"); model_info("conv1_weight");
@@ -311,7 +310,7 @@ u8 model_info(char* model_name){
printf("model.array.address is: 0X%X\r\n",_model -> array); printf("model.array.address is: 0X%X\r\n",_model -> array);
printf("model.maxlength is: %d\r\n",_model -> maxlength); printf("model.maxlength is: %d\r\n",_model -> maxlength);
printf("model.realength is: %d\r\n",_model -> realength); printf("model.realength is: %d\r\n",_model -> realength);
//if(strcmp(_model -> name, "data") == 0)printf("dataset is: %s\r\n",_model -> dname); 这函数有BUG，用model_dataset函数 //if(strcmp(_model -> name, "data") == 0)printf("dataset is: %s\r\n",_model -> dname); <EFBFBD>⺯<EFBFBD><EFBFBD><EFBFBD><EFBFBD>BUG<EFBFBD><EFBFBD><EFBFBD><EFBFBD>model_dataset<EFBFBD><EFBFBD><EFBFBD><EFBFBD>
return 1; return 1;
} }
return 0; return 0;
@@ -346,7 +345,7 @@ void model_init(){
conv1_weight.array = modelmym_init(conv1_weight.name); conv1_weight.array = modelmym_init(conv1_weight.name);
conv1_weight.maxlength = CONV1_WEIGHT_ARRSIZE; conv1_weight.maxlength = CONV1_WEIGHT_ARRSIZE;
conv1_weight.channel = 1; conv1_weight.channel = 1;
conv1_weight.num_kernels = 32; conv1_weight.num_kernels = CONV1_BIAS_ARRSIZE;
conv2_bias.name = "conv2_bias"; conv2_bias.name = "conv2_bias";
conv2_bias.array = modelmym_init(conv2_bias.name); conv2_bias.array = modelmym_init(conv2_bias.name);
@@ -356,7 +355,7 @@ void model_init(){
conv2_weight.array = modelmym_init(conv2_weight.name); conv2_weight.array = modelmym_init(conv2_weight.name);
conv2_weight.maxlength = CONV2_WEIGHT_ARRSIZE; conv2_weight.maxlength = CONV2_WEIGHT_ARRSIZE;
conv2_weight.channel = 32; conv2_weight.channel = 32;
conv2_weight.num_kernels = 64; conv2_weight.num_kernels = CONV2_BIAS_ARRSIZE;
conv3_bias.name = "conv3_bias"; conv3_bias.name = "conv3_bias";
conv3_bias.array = modelmym_init(conv3_bias.name); conv3_bias.array = modelmym_init(conv3_bias.name);
@@ -366,7 +365,7 @@ void model_init(){
conv3_weight.array = modelmym_init(conv3_weight.name); conv3_weight.array = modelmym_init(conv3_weight.name);
conv3_weight.maxlength = CONV3_WEIGHT_ARRSIZE; conv3_weight.maxlength = CONV3_WEIGHT_ARRSIZE;
conv3_weight.channel = 64; conv3_weight.channel = 64;
conv3_weight.num_kernels = 128; conv3_weight.num_kernels = CONV3_BIAS_ARRSIZE;
fc1_bias.name = "fc1_bias"; fc1_bias.name = "fc1_bias";
fc1_bias.array = modelmym_init(fc1_bias.name); fc1_bias.array = modelmym_init(fc1_bias.name);
@@ -383,7 +382,7 @@ void model_init(){
fc2_weight.name = "fc2_weight"; fc2_weight.name = "fc2_weight";
fc2_weight.array = modelmym_init(fc2_weight.name); fc2_weight.array = modelmym_init(fc2_weight.name);
fc2_weight.maxlength = FC2_WEIGHT_ARRSIZE; fc2_weight.maxlength = FC2_WEIGHT_ARRSIZE;
fc2_weight.num_kernels = FC2_WEIGHT_KERNELS; fc2_weight.num_kernels = FC2_BIAS_ARRSIZE;
data.name = "data"; data.name = "data";
data.array = modelmym_init(data.name); data.array = modelmym_init(data.name);

32
cnn_model.h
View File

@@ -1,8 +1,9 @@
#ifndef CNNMODEL_H_ #ifndef CNNMODEL_H_
#define CNNMODEL_H_ #define CNNMODEL_H_
#include "malloc.h" #include "malloc.h"
#include "debug.h" #include <math.h>
#include <stdio.h> #include <stdio.h>
#include <stdarg.h>
#include <stdlib.h> #include <stdlib.h>
#include <string.h> #include <string.h>
#include <stdint.h> #include <stdint.h>
@@ -13,11 +14,11 @@ typedef struct {
char* name; char* name;
char* dname; char* dname;
float* array; float* array;
u32 maxlength; uint32_t maxlength;
u32 realength; uint32_t realength;
u8 channel; uint8_t channel;
u8 num_kernels; uint8_t num_kernels;
} Model; } Model;
@@ -36,16 +37,7 @@ typedef struct {
#define FC2_BIAS_ARRSIZE (4) #define FC2_BIAS_ARRSIZE (4)
#define FC2_WEIGHT_ARRSIZE (4*128) //896 #define FC2_WEIGHT_ARRSIZE (4*128) //896
#define FC2_WEIGHT_KERNELS 4 //4个卷积核 #define DATA_ARRSIZE (1300000)
//1为跑原始数据 0为跑100矩阵
#define is1300000 1
#if is1300000
#define DATA_ARRSIZE (1300000)
#else
#define DATA_ARRSIZE (100 * 100)
#endif
@@ -64,11 +56,11 @@ extern Model data;
float* modelmym_init(char* model_name); float* modelmym_init(char* model_name);
u8 modelmym_free(char* model_name); uint8_t modelmym_free(char* model_name);
u8 model_write(char* model_name); uint8_t model_write(char* model_name);
u8 model_read(char* model_name, u32 start, u32 end, u32 gap); uint8_t model_read(char* model_name, uint32_t start, uint32_t end, uint32_t gap);
u8 model_switchdata(char* model_name); uint8_t model_switchdata(char* model_name);
u8 model_info(char* model_name); uint8_t model_info(char* model_name);
void* model(char* model_name); void* model(char* model_name);
void model_init(void); void model_init(void);

1300000
dataset/_data/split_3.txt Normal file
View File

File diff suppressed because it is too large Load Diff

1300000
dataset/_data/split_9.txt Normal file
View File

File diff suppressed because it is too large Load Diff

32
dataset/conv1_bias.txt Normal file
View File

@@ -0,0 +1,32 @@
-0.004631
-0.072553
-0.001263
-0.222981
0.005574
-0.430715
-0.312284
-0.161185
-0.151565
-0.060493
0.231827
-0.357113
-0.438887
0.198280
0.193470
-0.002525
0.179678
-0.308586
-0.112759
0.013523
-0.020758
-0.320550
0.258272
0.125773
-0.437729
0.151137
-0.316827
0.176797
-0.178724
0.223349
-0.279238
0.006178

288
dataset/conv1_weight.txt Normal file
View File

@@ -0,0 +1,288 @@
0.264396
0.012110
0.255190
-0.113358
0.184589
0.057568
0.191846
0.117760
-0.245086
-0.005652
0.013235
0.293202
-0.188579
0.014698
0.025587
-0.051998
0.127640
0.287372
-0.119986
-0.208348
0.278841
-0.066475
-0.081250
0.092410
-0.147814
0.047256
0.173974
-0.329876
-0.423296
-0.584921
-0.356580
-0.229025
-0.270002
-0.255455
0.316085
0.004567
0.028426
-0.334845
-0.003385
0.140219
0.062490
-0.118646
0.094126
0.255222
-0.107712
0.002896
-0.081413
-0.043819
0.240522
0.022667
0.231820
-0.123995
-0.079158
-0.108327
-0.027508
-0.109963
0.288606
-0.335862
-0.232113
0.018983
-0.193671
-0.184815
-0.126731
-0.178352
0.144650
0.175241
-0.073164
0.199136
-0.253890
0.004484
0.067553
0.093465
-0.234934
-0.018499
0.011184
0.112343
-0.244534
0.285953
0.256141
0.238208
-0.024591
-0.110319
-0.337829
-0.268405
-0.332071
-0.051145
0.120763
-0.024749
0.248383
-0.101862
-0.352574
-0.188207
-0.154518
-0.308905
-0.424403
-0.342620
0.047083
-0.204670
0.273949
0.185371
-0.183867
0.222050
-0.289864
-0.377326
-0.207679
-0.126656
0.228524
-0.135648
0.092356
-0.157213
0.134332
0.084983
-0.017297
-0.296890
-0.262142
-0.014616
0.111652
-0.087568
0.171141
-0.299918
0.290131
-0.128769
-0.010415
0.076719
0.221479
-0.157894
0.068890
0.009761
-0.258660
-0.186225
0.254293
0.106144
-0.124818
-0.352556
0.231044
0.277191
-0.140471
-0.176610
0.164834
-0.223433
0.117251
-0.090913
-0.154972
-0.073200
0.225833
-0.121053
-0.519649
-0.030060
0.188377
-0.082648
0.023897
-0.181276
-0.320161
-0.050713
-0.041318
0.074467
0.171705
-0.126269
0.033716
-0.183754
-0.362817
-0.192327
-0.358436
0.258644
0.221722
0.208344
-0.113581
-0.216356
-0.303875
0.022764
0.062494
-0.168767
-0.236584
0.388117
-0.067814
0.008403
0.050595
-0.108519
-0.262208
0.319813
-0.006849
-0.024721
0.007692
0.137184
0.061543
-0.048481
-0.166637
-0.211058
-0.306806
-0.057978
-0.096503
0.288373
0.151038
-0.333258
-0.186567
0.161674
0.262806
0.237353
-0.361782
-0.131101
-0.373160
-0.334982
-0.396515
-0.323307
-0.202191
-0.373635
0.046574
0.177337
-0.233079
0.197044
-0.324467
0.273250
-0.320781
0.233871
0.165834
-0.151728
-0.184531
0.166097
-0.040754
0.147207
0.131238
0.264796
0.291155
-0.200281
0.195431
0.128389
0.181558
-0.252029
0.181836
0.083001
-0.180665
-0.252472
-0.259574
-0.150548
-0.247819
-0.253767
-0.109555
0.158506
-0.197960
0.196803
0.252319
-0.030197
-0.055893
0.254212
0.229382
-0.207656
0.037787
0.028231
-0.347745
-0.275953
-0.166275
-0.229568
0.267331
-0.257930
0.010068
0.209660
0.150940
-0.060589
-0.319203
0.035770
0.092334
-0.230210
0.245727
-0.068963
0.086573
-0.061331
0.122903
0.080436
0.119416
-0.024226
-0.188355
-0.234652
0.152925
0.201060
0.063903
0.123116
0.042292
-0.172743
0.206093
0.406446
-0.190303
-0.234680
-0.100517
-0.102583
0.154145
-0.134558
0.108717
0.010376

64
dataset/conv2_bias.txt Normal file
View File

@@ -0,0 +1,64 @@
-0.005641
0.045337
0.022688
0.022603
-0.003216
-0.030439
-0.141888
0.079657
-0.013354
-0.038644
-0.059302
-0.031407
-0.038632
0.039465
0.048158
-0.046950
-0.032638
-0.062939
-0.034895
0.013741
-0.037496
-0.019306
-0.007166
-0.045438
-0.017623
-0.029193
0.109553
-0.093409
0.000137
-0.009409
0.028830
0.002159
0.052856
-0.064427
-0.015227
0.012225
0.000874
-0.005634
0.036257
-0.054466
-0.003907
-0.143422
0.013692
-0.030289
-0.038533
-0.112211
0.034781
-0.008203
-0.054083
-0.016784
0.033634
-0.088775
0.017148
0.021540
0.015659
0.011183
-0.001851
-0.031995
-0.080671
0.051828
0.059321
-0.025378
-0.054386
-0.016398

18432
dataset/conv2_weight.txt Normal file
View File

File diff suppressed because it is too large Load Diff

128
dataset/conv3_bias.txt Normal file
View File

@@ -0,0 +1,128 @@
0.080931
-0.005371
-0.036383
-0.045386
-0.054329
0.108908
-0.031541
0.028897
-0.060491
-0.020571
-0.018694
-0.025861
-0.030109
-0.052287
-0.096520
-0.006112
0.012281
-0.040138
-0.026996
-0.028355
0.023113
-0.018347
0.021012
-0.024404
-0.020810
-0.033943
-0.031695
-0.002791
-0.005490
0.052321
-0.037635
-0.028975
-0.072508
-0.010565
0.089627
0.051853
-0.049895
-0.028532
0.120244
-0.065720
-0.033159
0.008072
-0.022876
0.002690
0.028910
0.019758
-0.027900
-0.018999
-0.040230
0.012506
-0.044383
-0.028107
-0.035639
-0.008712
-0.042572
-0.007128
-0.026243
-0.047105
-0.071900
0.004547
-0.037700
-0.033511
-0.045946
-0.062072
-0.063012
0.037648
0.014312
0.006599
-0.023144
-0.023018
0.013904
-0.052499
0.017902
-0.040357
-0.098850
0.011475
0.070717
-0.013301
0.003174
-0.030168
-0.034066
-0.013226
-0.023046
-0.030439
-0.026275
0.031229
-0.053149
-0.067014
0.003809
0.001963
0.023235
-0.000045
-0.027270
-0.007391
-0.012155
-0.003669
-0.041546
0.016943
0.022902
-0.045985
-0.053092
0.021105
-0.032606
-0.028661
-0.048068
-0.042106
-0.017811
-0.019530
-0.028386
-0.047340
0.015859
-0.040411
-0.006513
0.007619
0.012492
0.011398
0.030425
-0.049798
-0.052375
-0.017007
-0.002935
0.080796
-0.064879
0.012029
0.002495
-0.053315
0.021447
-0.037960

73728
dataset/conv3_weight.txt Normal file
View File

File diff suppressed because it is too large Load Diff

128
dataset/fc1_bias.txt Normal file
View File

@@ -0,0 +1,128 @@
-0.003303
-0.017162
-0.012434
-0.042197
-0.033317
-0.043120
0.000561
-0.011677
-0.000354
-0.008174
-0.032653
-0.000064
-0.007428
0.006177
-0.005123
-0.042697
-0.007818
-0.025033
0.017732
-0.065127
-0.036435
-0.038140
0.016812
-0.012877
-0.035449
-0.035985
0.000884
-0.026337
-0.003007
0.001144
-0.022054
0.032519
-0.008320
-0.004585
0.015589
-0.028719
-0.021946
-0.029944
-0.026863
-0.012840
-0.002333
-0.078329
-0.016042
-0.001515
-0.007396
0.017858
-0.014112
-0.040561
0.000165
-0.032233
0.014843
-0.055889
-0.032679
-0.002448
-0.023142
-0.031932
-0.061502
-0.011828
0.054338
0.025108
-0.036360
-0.014974
-0.005564
0.006727
-0.001447
-0.011361
-0.032685
-0.042424
0.028065
-0.010904
-0.018815
0.019042
-0.014854
-0.016071
-0.028355
-0.010906
-0.002750
-0.004453
-0.036549
0.008669
-0.031032
-0.004035
-0.004741
0.009571
-0.013948
-0.024394
-0.014121
-0.000113
0.030506
-0.020288
0.024610
-0.008582
0.029013
-0.003862
0.005664
-0.034357
0.015551
-0.007669
0.004395
-0.022629
-0.005918
-0.037288
0.000774
-0.049784
-0.010008
0.006821
-0.006700
0.000237
-0.005140
-0.010332
0.035143
-0.009957
-0.010226
-0.004023
-0.021022
0.009519
0.002547
-0.019027
-0.007719
-0.005774
-0.012114
0.008235
-0.014013
-0.001604
0.003618
-0.007368
0.010152
-0.027849

2359296
dataset/fc1_weight.txt Normal file
View File

File diff suppressed because it is too large Load Diff

4
dataset/fc2_bias.txt Normal file
View File

@@ -0,0 +1,4 @@
-0.039707
-0.097658
0.063971
0.022999

512
dataset/fc2_weight.txt Normal file
View File

@@ -0,0 +1,512 @@
-0.058931
0.082741
0.021586
-0.026039
-0.056528
0.057236
-0.030001
-0.048260
-0.062991
-0.012605
-0.008995
0.075796
0.001260
0.070238
-0.038573
-0.010643
-0.061193
-0.055042
0.076177
-0.079882
0.052359
0.034187
0.046499
0.045794
-0.020017
0.001001
-0.040188
-0.079907
-0.022438
0.077808
-0.026346
0.016867
-0.062112
0.064634
-0.019558
-0.020223
0.074465
-0.026598
0.011714
-0.064458
-0.059716
-0.069215
-0.052678
0.051622
0.010130
0.075898
-0.060538
-0.024788
-0.023184
-0.051868
0.158896
0.028924
-0.009646
-0.042931
-0.054899
-0.068251
-0.013731
-0.074485
0.034882
0.048279
-0.028229
-0.033060
-0.042540
0.069897
-0.015747
-0.062755
0.050984
-0.058438
-0.000549
0.037112
-0.008702
0.052537
-0.032729
0.085430
0.005522
0.036936
-0.035021
-0.047576
0.021966
0.049943
0.004276
-0.002239
-0.025785
0.082176
0.004891
-0.070603
-0.061674
0.076097
0.054244
-0.121986
-0.098606
0.020275
0.089327
-0.065931
0.035219
-0.030433
0.016048
-0.082111
-0.005972
-0.033918
-0.045327
-0.055175
-0.021159
-0.045618
-0.045335
0.080924
-0.043546
-0.011036
0.033512
0.030838
-0.125617
0.027708
-0.081700
0.065922
0.051387
0.084328
0.047665
-0.026571
-0.035575
-0.024575
0.046240
0.077493
0.061416
-0.060857
0.001534
-0.071932
0.028963
-0.075439
-0.017644
0.054737
-0.003758
0.052111
0.011470
0.055459
-0.081584
-0.002910
-0.001614
0.000437
-0.010231
-0.080488
-0.063019
0.034179
-0.069775
0.043572
-0.049781
0.072616
0.047185
0.030650
0.082894
0.046069
-0.027637
0.023121
0.038936
0.052421
-0.069802
0.077418
-0.046277
-0.008017
0.001110
0.032984
-0.056455
-0.036477
-0.012413
-0.005958
-0.032892
-0.021581
0.007284
0.061327
-0.051783
-0.007153
-0.034280
0.085166
0.074945
0.028743
0.059292
0.034775
-0.081315
0.087160
-0.116786
0.049679
0.054562
-0.052033
0.037010
0.039532
0.042317
-0.055750
0.060417
-0.009291
0.044491
0.046893
-0.034242
-0.014553
-0.054940
-0.081220
0.085371
-0.014649
0.024209
0.089537
0.016395
-0.023147
0.058076
0.070385
0.006581
-0.001097
0.004920
0.048413
0.068944
-0.047064
-0.022705
-0.037674
-0.048809
0.040081
0.065193
0.043707
0.022925
0.066769
-0.012898
-0.077912
-0.052976
-0.081205
-0.023203
-0.015376
-0.003481
-0.002040
-0.040693
-0.030009
-0.038227
0.033057
-0.044845
0.066074
-0.025737
0.093399
0.059980
-0.061411
0.068019
0.067066
-0.037720
0.035344
-0.099108
0.057824
0.010723
0.073095
0.064890
-0.069253
-0.035479
0.014207
0.033675
0.036457
0.037335
-0.056124
0.070666
-0.016997
-0.042993
0.019509
0.010573
-0.073225
0.010903
0.021510
0.046624
0.016918
-0.034955
-0.004001
-0.071385
0.052190
0.054234
0.028797
-0.016552
-0.049426
-0.008711
-0.003803
0.060791
-0.065849
0.068911
0.072407
0.052717
-0.037600
-0.029691
0.041340
-0.003574
0.028065
-0.056100
-0.039326
0.015791
0.011882
0.045606
-0.016313
-0.063804
-0.023912
-0.007011
0.080493
-0.011968
-0.048865
0.068336
0.063640
-0.018984
0.016588
-0.080367
-0.076863
0.069094
0.075082
0.072882
-0.018639
-0.034967
-0.034542
-0.057999
0.033039
-0.045209
0.023652
-0.051687
-0.077231
0.046792
0.048312
0.084837
0.036035
0.003179
-0.019603
-0.036890
-0.063492
-0.045074
-0.083299
0.024020
0.047446
-0.078694
-0.071320
0.062962
0.072505
-0.048483
0.005049
0.013660
-0.064950
-0.032050
-0.026682
0.072182
0.080960
-0.044191
-0.041733
0.073146
-0.064979
0.078111
-0.039892
-0.002631
0.024524
0.063003
0.063028
0.022161
0.036528
0.005402
0.017228
0.012027
-0.017038
0.072777
0.000782
-0.042552
0.014487
0.000573
0.032583
0.048010
0.043475
-0.041002
0.024025
0.073455
0.013298
0.074902
0.064317
0.045918
-0.005138
0.003448
0.078688
0.048412
0.047301
0.061309
-0.015053
0.003955
-0.067288
0.076958
0.033469
-0.058862
0.018029
0.078768
0.039408
-0.033360
-0.022925
-0.119977
-0.025087
-0.068750
0.035327
0.073855
0.047400
-0.075196
0.010550
-0.046624
0.029754
-0.057842
-0.052033
0.017090
0.019803
0.015316
-0.043967
0.015191
-0.061190
-0.021009
0.074461
0.034960
-0.051407
-0.044240
0.029745
-0.033547
-0.027608
-0.006036
0.030190
-0.007352
-0.065457
0.040382
0.001965
-0.033753
0.075904
0.003190
0.088717
0.082109
-0.039526
-0.002976
-0.028223
-0.028789
-0.069433
0.030751
-0.041434
0.007254
0.040745
0.063343
-0.102280
-0.021423
-0.042879
0.065601
0.064359
-0.098568
-0.014232
-0.023211
-0.014910
0.090127
-0.109398
-0.166171
-0.019305
0.074800
-0.038058
0.012383
0.078751
0.045327
-0.020477
-0.010302
-0.045507
0.033407
-0.035723
0.075363
-0.055598
0.062439
0.006332
-0.011081
-0.023165
-0.097509
0.051888
0.005613
0.042150
0.026668
-0.085304
0.069341
-0.081594
-0.054077
0.075259
-0.074176
-0.072364
-0.071474
-0.015352
-0.084154
0.034522
0.073654
-0.020917
-0.002134
-0.049188
-0.054401
-0.064645
-0.093849
0.072262
0.071346
-0.067045
-0.009426
-0.000583
0.070907
-0.070580
0.029851
0.061377
-0.040588
0.056680
-0.075749
0.076125
0.009514
0.015952
0.092470
-0.072424
0.045863
0.036735
-0.078604
0.061958
-0.081167
0.005184
0.022731
-0.023145
-0.013089
-0.000382
-0.053056
-0.019507
0.018815
-0.035313

27
debug.c
View File

@@ -1,27 +0,0 @@
#include "debug.h"
u8 _DEBUG = 0;
void DEBUG(){
_DEBUG = !_DEBUG;
printf("DEBUG IS: %s",(_DEBUG ? "ON" : "OFF"));
}
void DEBUG_PRINTF(const char *fmt, ...) {
if(_DEBUG){
va_list args;
va_start(args, fmt);
vprintf(fmt, args);
va_end(args);
}
}
void SDRAM_USED(){
}

19
debug.h
View File

@@ -1,19 +0,0 @@
#ifndef DEBUG_H_
#define DEBUG_H_
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include "malloc.h"
typedef uint32_t u32;
typedef uint16_t u16;
typedef uint8_t u8;
void DEBUG(void);
void DEBUG_PRINTF(const char *fmt, ...);
void SDRAM_USED(void);
#endif

102
main.c
View File

@@ -5,108 +5,10 @@
#include "cnn_model.h" #include "cnn_model.h"
#include "cnn.h" #include "cnn.h"
void run_dataset(){
char* modelusearr_origin[] = {
"C1autosave00095_right_new_2",
"C1autosave00096_right_new_2",
"C1autosave00097_right_new_2",
"C1autosave00098_right_new_2",
"C1autosave00099_right_new_2",
"C1autosave00100_right_new_2",
"C1autosave00101_right_new_2",
"C1autosave00102_right_new_2",
"C1autosave00103_right_new_2",
"C1autosave00104_right_new_2",
"C1autosave00105_right_new_2",
"C1autosave00106_right_new_2",
"C1autosave00107_right_new_2",
"C1autosave00108_right_new_2",
"C1autosave00109_right_new_2",
"C1autosave00110_right_new_2",
"C1autosave00111_right_new_2",
"C1autosave00112_right_new_2",
"C1autosave00113_right_new_2",
"C1autosave00114_right_new_2",
"C1autosave00115_right_new_2",
"C1autosave00116_right_new_2",
"C1autosave00117_right_new_2",
"C1autosave00118_right_new_2",
"C1autosave00119_right_new_2",
"C1autosave00120_right_new_2",
"C1autosave00121_right_new_2",
"C1autosave00122_right_new_2",
"C1autosave00123_right_new_2",
"C1autosave00124_right_new_2",
"filtered_C1autosave00011_right_new",
"filtered_C1autosave00015_right_new",
"filtered_C1autosave00043_right_new",
"filtered_C1autosave00067_right_new",
"filtered_C1autosave00090_right_new",
"filtered_C1autosave00106_right_new",
"filtered_C1autosave00118_right_new",
"filtered_C1autosave00007_right_new",
"filtered_C1autosave00035_right_new",
"filtered_C1autosave00036_right_new",
"filtered_C1autosave00040_right_new",
"filtered_C1autosave00053_right_new",
"filtered_C1autosave00061_right_new",
"filtered_C1autosave00074_right_new",
"filtered_C1autosave00077_right_new",
"filtered_C1autosave00080_right_new",
"filtered_C1autosave00085_right_new",
"filtered_C1autosave00098_right_new",
"filtered_C1autosave00100_right_new",
"filtered_C1autosave00104_right_new",
"filtered_C1autosave00122_right_new",
"filtered_C1autosave00124_right_new",
"filtered_C1autosave00108_right_new",
"filtered_C1autosave00004_right_new",
"filtered_C1autosave00039_right_new",
"filtered_C1autosave00062_right_new",
};
char* modelusearr_output[] = {
"split_1",
"split_2",
"split_3",
"split_4",
"split_5",
"split_6",
"split_7",
"split_8",
"split_9",
"split_10",
};
/* 可用数据集如下
* modelusearr_origin
* modelusearr_output
*/
#define modelusearr modelusearr_output
for(int a=0;a<(sizeof(modelusearr) / sizeof(modelusearr[0]));a++){
SDRAM_USED();
model_switchdata(modelusearr[a]);
cnn_run();
}
printf("\r\n运行完成\r\n");
}
int main(){ int main(){
model_init(); model_init();
model_write("all"); model_write("all");
model_switchdata("split_3");
#if is1300000
run_dataset();
#else
model_switchdata("EEC30B3F 2024-11-13 11-49-24");
cnn_run(); cnn_run();
#endif printf("\r\n<EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>\r\n");
DEBUG_PRINTF("\r\nEnd结束");
} }