// Gabor_Fiter_lh_3.cpp
// 2016/07/23 by marsee
// 2016/07/25 : 右白線検出用のGabor Filterを追加して、右左の白線を指定するRorL 引数を追加
// 2016/07/27 : 右白線検出用配列と左白線検出用配列を統合
// 2016/08/29 : 1回目左白線検出、2回目右白線検出のモードを追加
// 2022/12/14 : Vitis HLS用に書き直した
// 2022/12/27 : 引数にrow_size, col_size, functionを追加
// functionの値で元画像をスルーとガボール・フィルタ処理を選択できるようにした
// 2022/12/30 : user=1を待つdo while()文とmain()中のif文を削除した
//
#include <stdio.h>
#include <string.h>
#include <stdint.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <ap_axi_sdata.h>
#include "Gabor_Filter_lh_3.h"
int conv_rgb2y(int rgb);
int Gabor_Filter_lh_3(hls::stream<ap_axis<32,1,1,1> >& ins,
hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL,
int32_t row_size, int32_t col_size, int32_t function){
#pragma HLS INTERFACE mode=s_axilite port=function
#pragma HLS INTERFACE mode=s_axilite port=col_size
#pragma HLS INTERFACE mode=s_axilite port=row_size
#pragma HLS INTERFACE mode=ap_none port=RorL
#pragma HLS INTERFACE mode=axis register_mode=both port=ins register
#pragma HLS INTERFACE mode=axis register_mode=both port=outs register
#pragma HLS INTERFACE mode=s_axilite port=return
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> gabor;
ap_int<32> line_buf[ARRAY_SIZE-1][1920]; // Up to HD resolution
#pragma HLS array_partition variable=line_buf block factor=9 dim=1
ap_int<32> pix_mat[ARRAY_SIZE][ARRAY_SIZE];
int gray_pix, val, i, j, x, y, k;
LOOP_LR: for (k=0; k<2; k++){
LOOP_Y: for (y=0; y<row_size; y++){
#pragma HLS LOOP_TRIPCOUNT avg=600 max=1080 min=48
LOOP_X: for (x=0; x<col_size; x++){
#pragma HLS LOOP_TRIPCOUNT avg=800 max=1920 min=64
#pragma HLS PIPELINE II=1
ins >> pix; // AXI4-Stream からの入力
LOOP_PIX_MAT_K: for(int k=0; k<ARRAY_SIZE; k++){
LOOP_PIX_MAT_M: for(int m=0; m<ARRAY_SIZE-1; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
for (i=0; i<ARRAY_SIZE-1; i++){
pix_mat[i][ARRAY_SIZE-1] = line_buf[i][x];
}
ap_int<32> y_val = conv_rgb2y(pix.data);
pix_mat[ARRAY_SIZE-1][ARRAY_SIZE-1] = y_val;
for (i=0; i<ARRAY_SIZE-2; i++){ // 行の入れ替え
line_buf[i][x] = line_buf[i+1][x];
}
line_buf[ARRAY_SIZE-2][x] = y_val;
// 使用する配列を決定する
int ano;
switch (RorL){
case LEFT_WEIGHT :
ano = LEFT_WEIGHT;
break;
case RIGHT_WEIGHT :
ano = RIGHT_WEIGHT;
break;
case L_R_WEIGHT :
if (k == 0)
ano = LEFT_WEIGHT;
else
ano = RIGHT_WEIGHT;
break;
default :
ano = L_R_WEIGHT;
break;
}
// Gabor filter の演算
for (j=0, val=0; j<ARRAY_SIZE; j++){
for (i=0; i<ARRAY_SIZE; i++){
val += gabor_weight[ano][j][i] * pix_mat[j][i];
}
}
val = val/256; // 256倍してあるので、1/256して戻す
if (val<0)
//val = -val; // 絶対値
val = 0; // マイナスの値を0に丸める
else if (val>255)
val = 255;
// Gabor filter・データの書き込み
if(function == Gabor_Filter)
gabor.data = (val<<16)+(val<<8)+val;
else
gabor.data = pix.data;
// 最初のARRAY_SIZE-1行とその他の行の最初のARRAY_SIZE-1列は無効データなので0とする
if (x<(ARRAY_SIZE-1) || y<(ARRAY_SIZE-1)){
if(function == Gabor_Filter)
gabor.data = 0;
}
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
gabor.user = 1;
else
gabor.user = 0;
if (x == (col_size-1)) // 行の最後で TLAST をアサートする
gabor.last = 1;
else
gabor.last = 0;
outs << gabor; // AXI4-Stream へ出力
}
}
}
return(0);
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
// Gabor_filter_lh_3.h
// 2016/07/24
// 2016/07/25 : 右白線検出用のGabor Filterの重みを追加
// 2016/07/27 : 右白線検出用配列と左白線検出用配列を統合
// 2016/08/29 : 1回目左白線検出、2回目右白線検出のモードを追加
// 2022/12/27 : 引数にrow_size, col_size, functionを追加
// functionの値で元画像をスルーとガボール・フィルタ処理を選択できるようにした
//
#ifndef __Gabor_filter_lh_H__
#define __Gabor_filter_lh_H__
#define ARRAY_SIZE 9
#define RIGHT_WEIGHT 1
#define LEFT_WEIGHT 0
#define L_R_WEIGHT 2
#define Through 0
#define Gabor_Filter 1
const int gabor_weight[2][ARRAY_SIZE][ARRAY_SIZE] = { // 左白線検出用＋右白線検出用
{
{0,-3,-10,-16,-7,7,10,5,1},
{-3,-15,-27,-11,32,50,29,5,-2},
{-11,-24,-5,73,135,95,16,-17,-10},
{-11,4,85,187,160,14,-72,-52,-14},
{4,51,135,137,-18,-159,-136,-45,-2},
{16,50,59,-39,-179,-185,-73,3,13},
{10,12,-25,-104,-131,-60,15,27,11},
{1,-7,-31,-48,-24,18,29,14,3},
{-1,-5,-9,-4,10,16,10,2,-1}
},
{
{1,5,7,1,-12,-17,-10,-3,0},
{1,11,33,45,21,-16,-27,-15,-4},
{-8,-5,35,107,131,69,-2,-21,-11},
{-17,-47,-51,40,169,187,93,13,-7},
{-8,-54,-134,-147,-18,123,130,58,11},
{9,-6,-82,-185,-187,-65,36,44,17},
{11,24,12,-55,-125,-112,-43,1,7},
{3,14,30,23,-13,-41,-33,-12,-1},
{0,2,9,17,14,1,-6,-5,-2}
}
};
const float gabor_fweight[2][ARRAY_SIZE][ARRAY_SIZE] = { // 左白線検出用＋右白線検出用(float)
{
{0.001282,-0.009914,-0.04062,-0.060586,-0.027574,0.026072,0.038427,0.018191,0.003056},
{-0.012155,-0.057448,-0.104645,-0.042953,0.123263,0.197238,0.114451,0.020448,-0.007239},
{-0.042252,-0.093065,-0.018911,0.285374,0.525746,0.372687,0.060734,-0.064748,-0.040465},
{-0.042261,0.015673,0.332798,0.728763,0.625046,0.053591,-0.283076,-0.203293,-0.05608},
{0.017342,0.198305,0.52554,0.535526,-0.069756,-0.622839,-0.531089,-0.177366,-0.006367},
{0.060866,0.19708,0.231032,-0.154219,-0.699885,-0.721808,-0.286707,0.013004,0.049249},
{0.038379,0.04877,-0.098477,-0.404993,-0.510165,-0.233566,0.057894,0.104366,0.041887},
{0.0047,-0.0278,-0.121277,-0.187262,-0.093276,0.070512,0.113857,0.055799,0.009976},
{-0.003798,-0.01885,-0.035607,-0.01709,0.037692,0.064268,0.038606,0.007536,-0.002133}
},
{
{0.005562,0.018882,0.028293,0.004499,-0.044995,-0.064838,-0.039469,-0.009822,0.000815},
{0.002294,0.04108,0.127023,0.175094,0.083025,-0.063755,-0.106402,-0.057798,-0.01406},
{-0.031269,-0.021096,0.135641,0.417286,0.512467,0.269946,-0.008354,-0.082091,-0.041357},
{-0.066348,-0.184919,-0.197802,0.15614,0.65976,0.728616,0.361674,0.052074,-0.027152},
{-0.031146,-0.211178,-0.523777,-0.573856,-0.069756,0.480311,0.506451,0.225223,0.041031},
{0.035552,-0.023892,-0.320104,-0.723563,-0.728735,-0.253689,0.1391,0.170625,0.067723},
{0.04216,0.094939,0.047511,-0.216623,-0.488075,-0.437898,-0.168739,0.003336,0.027009},
{0.012112,0.056596,0.115239,0.090332,-0.05076,-0.158403,-0.127847,-0.046375,-0.004918},
{-0.00168,0.007437,0.036985,0.067021,0.053689,0.004977,-0.02365,-0.018248,-0.005928}
}
};
#endif
// Gabor_Filter_lh_3.cpp
// 2016/07/23 by marsee
// 2016/07/25 : 右白線検出用のGabor Filterを追加して、右左の白線を指定するRorL 引数を追加
// 2016/07/27 : 右白線検出用配列と左白線検出用配列を統合
// 2016/08/29 : 1回目左白線検出、2回目右白線検出のモードを追加
// 2022/12/14 : Vitis HLS用に書き直した
// 2022/12/27 : 引数にrow_size, col_size, functionを追加
// functionの値で元画像をスルーとガボール・フィルタ処理を選択できるようにした
//
#include <stdio.h>
#include <string.h>
#include <stdint.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <ap_axi_sdata.h>
#include "Gabor_Filter_lh_3.h"
int conv_rgb2y(int rgb);
int Gabor_Filter_lh_3(hls::stream<ap_axis<32,1,1,1> >& ins,
hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL,
int32_t row_size, int32_t col_size, int32_t function){
#pragma HLS INTERFACE mode=s_axilite port=function
#pragma HLS INTERFACE mode=s_axilite port=col_size
#pragma HLS INTERFACE mode=s_axilite port=row_size
#pragma HLS INTERFACE mode=ap_none port=RorL
#pragma HLS INTERFACE mode=axis register_mode=both port=ins register
#pragma HLS INTERFACE mode=axis register_mode=both port=outs register
#pragma HLS INTERFACE mode=s_axilite port=return
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> gabor;
ap_int<32> line_buf[ARRAY_SIZE-1][1920]; // Up to HD resolution
#pragma HLS array_partition variable=line_buf block factor=9 dim=1
ap_int<32> pix_mat[ARRAY_SIZE][ARRAY_SIZE];
int gray_pix, val, i, j, x, y, k;
LOOP_WAIT_USER: do {
#pragma HLS LOOP_TRIPCOUNT avg=1 max=1 min=1
// user が 1になった時にフレームがスタートする
ins >> pix;
} while(pix.user == 0);
LOOP_LR: for (k=0; k<2; k++){
LOOP_Y: for (y=0; y<row_size; y++){
#pragma HLS LOOP_TRIPCOUNT avg=600 max=1080 min=48
LOOP_X: for (x=0; x<col_size; x++){
#pragma HLS LOOP_TRIPCOUNT avg=800 max=1920 min=64
#pragma HLS PIPELINE II=1
if (!(k==0 && x==0 && y==0)) // 最初の入力はすでに入力されている
ins >> pix; // AXI4-Stream からの入力
LOOP_PIX_MAT_K: for(int k=0; k<ARRAY_SIZE; k++){
LOOP_PIX_MAT_M: for(int m=0; m<ARRAY_SIZE-1; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
for (i=0; i<ARRAY_SIZE-1; i++){
pix_mat[i][ARRAY_SIZE-1] = line_buf[i][x];
}
ap_int<32> y_val = conv_rgb2y(pix.data);
pix_mat[ARRAY_SIZE-1][ARRAY_SIZE-1] = y_val;
for (i=0; i<ARRAY_SIZE-2; i++){ // 行の入れ替え
line_buf[i][x] = line_buf[i+1][x];
}
line_buf[ARRAY_SIZE-2][x] = y_val;
// 使用する配列を決定する
int ano;
switch (RorL){
case LEFT_WEIGHT :
ano = LEFT_WEIGHT;
break;
case RIGHT_WEIGHT :
ano = RIGHT_WEIGHT;
break;
case L_R_WEIGHT :
if (k == 0)
ano = LEFT_WEIGHT;
else
ano = RIGHT_WEIGHT;
break;
default :
ano = L_R_WEIGHT;
break;
}
// Gabor filter の演算
for (j=0, val=0; j<ARRAY_SIZE; j++){
for (i=0; i<ARRAY_SIZE; i++){
val += gabor_weight[ano][j][i] * pix_mat[j][i];
}
}
val = val/256; // 256倍してあるので、1/256して戻す
if (val<0)
//val = -val; // 絶対値
val = 0; // マイナスの値を0に丸める
else if (val>255)
val = 255;
// Gabor filter・データの書き込み
if(function == Gabor_Filter)
gabor.data = (val<<16)+(val<<8)+val;
else
gabor.data = pix.data;
// 最初のARRAY_SIZE-1行とその他の行の最初のARRAY_SIZE-1列は無効データなので0とする
if (x<(ARRAY_SIZE-1) || y<(ARRAY_SIZE-1)){
if(function == Gabor_Filter)
gabor.data = 0;
}
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
gabor.user = 1;
else
gabor.user = 0;
if (x == (col_size-1)) // 行の最後で TLAST をアサートする
gabor.last = 1;
else
gabor.last = 0;
outs << gabor; // AXI4-Stream へ出力
}
}
}
return(0);
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
// Gabor_Filter_lh_3_tb.cpp
// 2016/07/24 by marsee
// 2016/07/25 : 右白線検出用のGabor Filterを追加して、右左の白線を指定するRorL 引数を追加
// 2022/12/14 : Vitis HLS用に書き直した
// 2022/12/27 : 引数にrow_size, col_size, functionを追加
// functionの値で元画像をスルーとガボール・フィルタ処理を選択できるようにした
//
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <iostream>
#include <fstream>
#include <math.h>
#include <ap_axi_sdata.h>
#include "Gabor_Filter_lh_3.h"
#include "bmp_header.h"
int Gabor_Filter_lh_3(hls::stream<ap_axis<32,1,1,1> >& ins,
hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL,
int32_t row_size, int32_t col_size, int32_t function);
int conv_rgb2y_soft(int rgb);
int Gabor_Filter_lh_3_soft(hls::stream<ap_axis<32,1,1,1> >& ins,
hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL,
int32_t row_size, int32_t col_size, int32_t function);
#define CLOCK_PERIOD 10
#define RIGHT_OR_LEFT L_R_WEIGHT
#define BMP_FILE_NAME "test.bmp"
int main()
{
using namespace std;
hls::stream<ap_axis<32,1,1,1> > ins, ins2;
hls::stream<ap_axis<32,1,1,1> > ins_soft;
hls::stream<ap_axis<32,1,1,1> > outs, outs2;
hls::stream<ap_axis<32,1,1,1> > outs_soft;
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> vals;
ap_axis<32,1,1,1> vals_soft;
int m_seq = 1; // M系列の値
int i, k;
int xor_shift;
BITMAPFILEHEADER bmpfhr; // BMPファイルのファイルヘッダ(for Read)
BITMAPINFOHEADER bmpihr; // BMPファイルのINFOヘッダ(for Read)
FILE *fbmpr, *fbmpw, *fbmpwf;
int *rd_bmp, *hw_gabor, *sw_gabor;
int blue, green, red;
ap_uint<2> r_l;
char fhname[100];
char fsname[100];
if ((fbmpr = fopen(BMP_FILE_NAME, "rb")) == NULL){ // test.bmp をオープン
fprintf(stderr, "Can't open test.bmp by binary read mode\n");
exit(1);
}
// bmpヘッダの読み出し
fread(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpr);
fread(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpr);
fread(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpr);
printf("bmpihr.biWidth = %d, bmpihr.biHeight = %d\n",(int)bmpihr.biWidth, (int)bmpihr.biHeight);
// ピクセルを入れるメモリをアロケートする
if ((rd_bmp =(int *)malloc(sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate rd_bmp memory\n");
exit(1);
}
if ((hw_gabor =(int *)malloc(2 * sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate hw_gabor memory\n");
exit(1);
}
if ((sw_gabor =(int *)malloc(2 * sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate hw_gabor memory\n");
exit(1);
}
// rd_bmp にBMPのピクセルを代入。その際に、行を逆転する必要がある
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = fgetc(fbmpr);
green = fgetc(fbmpr);
red = fgetc(fbmpr);
rd_bmp[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] = (blue & 0xff) | ((green & 0xff)<<8) | ((red & 0xff)<<16);
}
}
fclose(fbmpr);
// 2画面分を入力する
for(int k=0; k<2; k++){
for(int j=0; j < bmpihr.biHeight; j++){
for(i=0; i < bmpihr.biWidth; i++){
pix.data = (ap_int<32>)rd_bmp[(j*bmpihr.biWidth)+i];
if (j==0 && i==0) // 最初のデータの時に TUSER を 1 にする
pix.user = 1;
else
pix.user = 0;
if (i == bmpihr.biWidth-1) // 行の最後でTLASTをアサートする
pix.last = 1;
else
pix.last = 0;
ins << pix;
ins2 << pix;
ins_soft << pix;
}
}
}
r_l = (ap_uint<2>)RIGHT_OR_LEFT;
Gabor_Filter_lh_3(ins, outs, r_l, 48, 64, Gabor_Filter);
Gabor_Filter_lh_3_soft(ins_soft, outs_soft, r_l, 48, 64, Gabor_Filter);
// ハードウェアとソフトウェアのラプラシアン・フィルタの値のチェック
for(k=0; k<2; k++){
for(int j=0; j < bmpihr.biHeight; j++){
for(i=0; i < bmpihr.biWidth; i++){
outs >> vals;
outs_soft >> vals_soft;
ap_int<32> val = vals.data;
ap_int<32> val_soft = vals_soft.data;
hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+(j*bmpihr.biWidth)+i] = (int)val;
sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+(j*bmpihr.biWidth)+i] = (int)val_soft;
//printf("k=%d, j=%d, i=%d\n",k,j,i);
if ((double)pow((double)(val&0xff)-(val_soft&0xff),(double)2) > 4){ // 2乗誤差が4よりも大きい
printf("ERROR HW and SW results mismatch i = %ld, j = %ld, HW = %08x, SW = %08x\n", i, j, (int)val, (int)val_soft);
//return(1);
}
//if (vals.last)
//cout << "AXI-Stream is end" << endl;
}
}
}
cout << "Success HW and SW results match" << endl;
cout << endl;
// ハードウェアのラプラシアンフィルタの結果を temp_gabor0.bmp, temp_gabor1.bmp へ出力する
for(k=0; k<2; k++){
if(k == 0){
if ((fbmpw=fopen("temp_gabor0.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor0.bmp by binary write mode\n");
exit(1);
}
} else {
if ((fbmpw=fopen("temp_gabor1.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor1.bmp by binary write mode\n");
exit(1);
}
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpw);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpw);
fputc(green, fbmpw);
fputc(red, fbmpw);
}
}
fclose(fbmpw);
}
// 元画像を出力
Gabor_Filter_lh_3(ins2, outs2, r_l, 48, 64, Through);
// hw_gabor[]に画像を出力
for(k=0; k<2; k++){
for(int j=0; j < bmpihr.biHeight; j++){
for(i=0; i < bmpihr.biWidth; i++){
outs2 >> vals;
ap_int<32> val = vals.data;
hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+(j*bmpihr.biWidth)+i] = (int)val;
}
}
}
// ハードウェアのラプラシアンフィルタの結果を temp_gabor0.bmp, temp_gabor1.bmp へ出力する
for(k=0; k<2; k++){
if(k == 0){
if ((fbmpw=fopen("temp_org0.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_org0.bmp by binary write mode\n");
exit(1);
}
} else {
if ((fbmpw=fopen("temp_org1.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_org1.bmp by binary write mode\n");
exit(1);
}
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpw);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpw);
fputc(green, fbmpw);
fputc(red, fbmpw);
}
}
fclose(fbmpw);
}
// ソフトウェアのラプラシアンフィルタの結果を temp_gabor_float0.bmp, temp_gabor_float1.bmp へ出力する
for(k=0; k<2; k++){
if (k == 0){
if ((fbmpwf=fopen("temp_gabor_float0.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor_float0.bmp by binary write mode\n");
exit(1);
}
} else {
if ((fbmpwf=fopen("temp_gabor_float1.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor_float1.bmp by binary write mode\n");
exit(1);
}
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpwf);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpwf);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpwf);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpwf);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpwf);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpwf);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpwf);
fputc(green, fbmpwf);
fputc(red, fbmpwf);
}
}
fclose(fbmpwf);
}
free(rd_bmp);
free(hw_gabor);
return 0;
}
int Gabor_Filter_lh_3_soft(hls::stream<ap_axis<32,1,1,1> >& ins,
hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL,
int32_t row_size, int32_t col_size, int32_t function){
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> gabor;
ap_int<32> line_buf[ARRAY_SIZE-1][1920]; // Up to HD resolution
ap_int<32> pix_mat[ARRAY_SIZE][ARRAY_SIZE];
int gray_pix, val, i, j, x, y, k;
do {
// user が 1になった時にフレームがスタートする
ins >> pix;
} while(pix.user == 0);
for (k=0; k<2; k++){
for (y=0; y<row_size; y++){
for (x=0; x<col_size; x++){
if (!(k==0 && x==0 && y==0)) // 最初の入力はすでに入力されている
ins >> pix; // AXI4-Stream からの入力
LOOP_PIX_MAT_K: for(int k=0; k<ARRAY_SIZE; k++){
LOOP_PIX_MAT_M: for(int m=0; m<ARRAY_SIZE-1; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
for (i=0; i<ARRAY_SIZE-1; i++){
pix_mat[i][ARRAY_SIZE-1] = line_buf[i][x];
}
ap_int<32> y_val = conv_rgb2y_soft(pix.data);
pix_mat[ARRAY_SIZE-1][ARRAY_SIZE-1] = y_val;
for (i=0; i<ARRAY_SIZE-2; i++){ // 行の入れ替え
line_buf[i][x] = line_buf[i+1][x];
}
line_buf[ARRAY_SIZE-2][x] = y_val;
// 使用する配列を決定する
int ano;
switch (RorL){
case LEFT_WEIGHT :
ano = LEFT_WEIGHT;
break;
case RIGHT_WEIGHT :
ano = RIGHT_WEIGHT;
break;
case L_R_WEIGHT :
if (k == 0)
ano = LEFT_WEIGHT;
else
ano = RIGHT_WEIGHT;
break;
default :
ano = L_R_WEIGHT;
break;
}
// Gabor filter の演算
for (j=0, val=0; j<ARRAY_SIZE; j++){
for (i=0; i<ARRAY_SIZE; i++){
val += gabor_weight[ano][j][i] * pix_mat[j][i];
}
}
val = val/256; // 256倍してあるので、1/256して戻す
if (val<0)
//val = -val; // 絶対値
val = 0; // マイナスの値を0に丸める
else if (val>255)
val = 255;
// Gabor filter・データの書き込み
if(function == Gabor_Filter)
gabor.data = (val<<16)+(val<<8)+val;
else
gabor.data = pix.data;
// 最初のARRAY_SIZE-1行とその他の行の最初のARRAY_SIZE-1列は無効データなので0とする
if (x<(ARRAY_SIZE-1) || y<(ARRAY_SIZE-1)){
if(function == Gabor_Filter)
gabor.data = 0;
}
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
gabor.user = 1;
else
gabor.user = 0;
if (x == (col_size-1)) // 行の最後で TLAST をアサートする
gabor.last = 1;
else
gabor.last = 0;
outs << gabor; // AXI4-Stream へ出力
}
}
}
return(0);
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y_soft(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
// bmp_header.h
// BMP ファイルフォーマットから引用させて頂きました
// http://www.kk.iij4u.or.jp/~kondo/bmp/
//
// 2017/05/04 : takseiさんのご指摘によりintX_tを使った宣言に変更。takseiさんありがとうございました
// 変数の型のサイズの違いによってLinuxの６４ビット版では動作しなかったためです
// http://marsee101.blog19.fc2.com/blog-entry-3354.html#comment2808
//
#include <stdio.h>
#include <stdint.h>
// BITMAPFILEHEADER 14bytes
typedef struct tagBITMAPFILEHEADER {
uint16_t bfType;
uint32_t bfSize;
uint16_t bfReserved1;
uint16_t bfReserved2;
uint32_t bfOffBits;
} BITMAPFILEHEADER;
// BITMAPINFOHEADER 40bytes
typedef struct tagBITMAPINFOHEADER{
uint32_t biSize;
int32_t biWidth;
int32_t biHeight;
uint16_t biPlanes;
uint16_t biBitCount;
uint32_t biCompression;
uint32_t biSizeImage;
int32_t biXPixPerMeter;
int32_t biYPixPerMeter;
uint32_t biClrUsed;
uint32_t biClrImporant;
} BITMAPINFOHEADER;
typedef struct BMP24bitsFORMAT {
uint8_t blue;
uint8_t green;
uint8_t red;
} BMP24FORMAT;
// lap_filter_axis.h
// 2015/05/01
// 2022/12/26 : modified by marsee
// function value
#define Through 0
#define Lap_Filter 1
//
// lap_filter_axis.cpp
// by marsee
// 2015/05/01
// 2015/06/25 : 修正、ラプラシアンフィルタの値が青だけになっていたので、RGBに拡張した
// 2022/12/27 : 修正、Vitis_HLS 2022.2用に変更、引数にrow_size, col_size, functionを追加
// functionの値で元画像をスルーとラプラシアン・フィルタ処理を選択できるようにした
//
#include <stdio.h>
#include <stdint.h>
#include <string.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <ap_axi_sdata.h>
#include "lap_filter_axis.h"
int laplacian_fil(int x0y0, int x1y0, int x2y0, int x0y1, int x1y1, int x2y1, int x0y2, int x1y2, int x2y2);
int conv_rgb2y(int rgb);
int lap_filter_axis(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs,
uint32_t row_size, uint32_t col_size, uint32_t function){
#pragma HLS INTERFACE mode=s_axilite port=function
#pragma HLS INTERFACE mode=s_axilite port=col_size
#pragma HLS INTERFACE mode=s_axilite port=row_size
#pragma HLS INTERFACE mode=axis register_mode=both port=ins register
#pragma HLS INTERFACE mode=axis register_mode=both port=outs register
#pragma HLS INTERFACE mode=s_axilite port=return
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> lap;
unsigned int line_buf[2][1920];
#pragma HLS array_partition variable=line_buf block factor=2 dim=1
int pix_mat[3][3];
#pragma HLS array_partition variable=pix_mat complete
int lap_fil_val;
Loop1 : do { // user が 1になった時にフレームがスタートする
#pragma HLS LOOP_TRIPCOUNT min=1 max=1 avg=1
ins >> pix;
} while(pix.user == 0);
Loop2 : for (int y=0; y<row_size; y++){
#pragma HLS LOOP_TRIPCOUNT avg=600 max=1080 min=48
Loop3 : for (int x=0; x<col_size; x++){
#pragma HLS LOOP_TRIPCOUNT avg=800 max=1920 min=64
#pragma HLS PIPELINE II=1
if (!(x==0 && y==0)) // 最初の入力はすでに入力されている
ins >> pix; // AXI4-Stream からの入力
Loop4 : for (int k=0; k<3; k++){
Loop5 : for (int m=0; m<2; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
pix_mat[0][2] = line_buf[0][x];
pix_mat[1][2] = line_buf[1][x];
int y_val = conv_rgb2y(pix.data);
pix_mat[2][2] = y_val;
line_buf[0][x] = line_buf[1][x]; // 行の入れ替え
line_buf[1][x] = y_val;
lap_fil_val = laplacian_fil( pix_mat[0][0], pix_mat[0][1], pix_mat[0][2],
pix_mat[1][0], pix_mat[1][1], pix_mat[1][2],
pix_mat[2][0], pix_mat[2][1], pix_mat[2][2]);
if(function == Lap_Filter)
lap.data = (lap_fil_val<<16)+(lap_fil_val<<8)+lap_fil_val; // RGB同じ値を入れる
else
lap.data = pix.data; // 元画像
if (x<2 || y<2){ // 最初の2行とその他の行の最初の2列は無効データなので0とする
if(function == Lap_Filter)
lap.data = 0;
}
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
lap.user = 1;
else
lap.user = 0;
if (x == (col_size-1)) // 行の最後で TLAST をアサートする
lap.last = 1;
else
lap.last = 0;
outs << lap; // AXI4-Stream へ出力
}
}
return 0;
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
// ラプラシアンフィルタ
// x0y0 x1y0 x2y0 -1 -1 -1
// x0y1 x1y1 x2y1 -1 8 -1
// x0y2 x1y2 x2y2 -1 -1 -1
int laplacian_fil(int x0y0, int x1y0, int x2y0, int x0y1, int x1y1, int x2y1, int x0y2, int x1y2, int x2y2)
{
int y;
y = -x0y0 -x1y0 -x2y0 -x0y1 +8*x1y1 -x2y1 -x0y2 -x1y2 -x2y2;
if (y<0)
y = 0; // マイナスのエッジは0とする
// y = -y; // マイナスエッジを有効にする
else if (y>255)
y = 255;
return(y);
}
// lap_filter_axis_tb.cpp
// by marsee
// 2015/05/01
// 2015/08/17 : BMPファイルを読み書きするように変更した
// 2022/12/27 : 修正、Vitis_HLS 2022.2用に変更、引数にrow_size, col_size, functionを追加
// functionの値で元画像をスルーとラプラシアン・フィルタ処理を選択できるようにした
//
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <iostream>
#include <fstream>
#include <ap_axi_sdata.h>
#include "lap_filter_axis.h"
#include "bmp_header.h"
int lap_filter_axis(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs,
uint32_t row_size, uint32_t col_size, uint32_t function);
int laplacian_fil_soft(int x0y0, int x1y0, int x2y0, int x0y1, int x1y1, int x2y1, int x0y2, int x1y2, int x2y2);
int conv_rgb2y_soft(int rgb);
int lap_filter_axis_soft(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs, int width, int height);
int main()
{
using namespace std;
hls::stream<ap_axis<32,1,1,1> > ins, ins2;
hls::stream<ap_axis<32,1,1,1> > ins_soft;
hls::stream<ap_axis<32,1,1,1> > outs, outs2;
hls::stream<ap_axis<32,1,1,1> > outs_soft;
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> vals;
ap_axis<32,1,1,1> vals_soft;
BITMAPFILEHEADER bmpfhr; // BMPファイルのファイルヘッダ(for Read)
BITMAPINFOHEADER bmpihr; // BMPファイルのINFOヘッダ(for Read)
FILE *fbmpr, *fbmpw;
int *rd_bmp, *hw_lapd;
int blue, green, red;
if ((fbmpr = fopen("test.bmp", "rb")) == NULL){ // test.bmp をオープン
fprintf(stderr, "Can't open test.bmp by binary read mode\n");
exit(1);
}
// bmpヘッダの読み出し
fread(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpr);
fread(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpr);
fread(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpr);
// ピクセルを入れるメモリをアロケートする
if ((rd_bmp =(int *)malloc(sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate rd_bmp memory\n");
exit(1);
}
if ((hw_lapd =(int *)malloc(sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate hw_lapd memory\n");
exit(1);
}
// rd_bmp にBMPのピクセルを代入。その際に、行を逆転する必要がある
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = fgetc(fbmpr);
green = fgetc(fbmpr);
red = fgetc(fbmpr);
rd_bmp[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] = (blue & 0xff) | ((green & 0xff)<<8) | ((red & 0xff)<<16);
}
}
fclose(fbmpr);
// ins に入力データを用意する
for(int i=0; i<5; i++){ // dummy data
pix.user = 0;
pix.data = i;
ins << pix;
}
for(int j=0; j < bmpihr.biHeight; j++){
for(int i=0; i < bmpihr.biWidth; i++){
pix.data = (ap_int<32>)rd_bmp[(j*bmpihr.biWidth)+i];
if (j==0 && i==0) // 最初のデータの時に TUSER を 1 にする
pix.user = 1;
else
pix.user = 0;
if (i == bmpihr.biWidth-1) // 行の最後でTLASTをアサートする
pix.last = 1;
else
pix.last = 0;
ins << pix;
ins2 << pix;
ins_soft << pix;
}
}
lap_filter_axis(ins, outs, bmpihr.biHeight, bmpihr.biWidth, Lap_Filter);
lap_filter_axis_soft(ins_soft, outs_soft, bmpihr.biWidth, bmpihr.biHeight);
// ハードウェアとソフトウェアのラプラシアン・フィルタの値のチェック
cout << endl;
cout << "outs" << endl;
for(int j=0; j < bmpihr.biHeight; j++){
for(int i=0; i < bmpihr.biWidth; i++){
outs >> vals;
outs_soft >> vals_soft;
ap_int<32> val = vals.data;
ap_int<32> val_soft = vals_soft.data;
hw_lapd[(j*bmpihr.biWidth)+i] = (int)val;
if (val != val_soft){
printf("ERROR HW and SW results mismatch i = %ld, j = %ld, HW = %d, SW = %d\n", i, j, (int)val, (int)val_soft);
return(1);
}
}
}
cout << "Success HW and SW results match" << endl;
cout << endl;
// ハードウェアのラプラシアンフィルタの結果を temp_lap.bmp へ出力する
if ((fbmpw=fopen("temp_lap.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_lap.bmp by binary write mode\n");
exit(1);
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpw);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = hw_lapd[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (hw_lapd[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (hw_lapd[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpw);
fputc(green, fbmpw);
fputc(red, fbmpw);
}
}
fclose(fbmpw);
// 元画像のスルーパスをテスト
lap_filter_axis(ins2, outs2, bmpihr.biHeight, bmpihr.biWidth, Through);
for(int j=0; j < bmpihr.biHeight; j++){
for(int i=0; i < bmpihr.biWidth; i++){
outs2 >> vals;
ap_int<32> val = vals.data;
hw_lapd[(j*bmpihr.biWidth)+i] = (int)val;
}
}
// 元画像の出力結果を temp_org.bmp へ出力する
if ((fbmpw=fopen("temp_org.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_org.bmp by binary write mode\n");
exit(1);
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpw);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = hw_lapd[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (hw_lapd[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (hw_lapd[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpw);
fputc(green, fbmpw);
fputc(red, fbmpw);
}
}
fclose(fbmpw);
free(rd_bmp);
free(hw_lapd);
return 0;
}
int lap_filter_axis_soft(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs, int width, int height){
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> lap;
unsigned int **line_buf;
int pix_mat[3][3];
int lap_fil_val;
int i;
// line_buf の1次元目の配列をアロケートする
if ((line_buf =(unsigned int **)malloc(sizeof(unsigned int *) * 2)) == NULL){
fprintf(stderr, "Can't allocate line_buf[3][]\n");
exit(1);
}
// メモリをアロケートする
for (i=0; i<2; i++){
if ((line_buf[i]=(unsigned int *)malloc(sizeof(unsigned int) * width)) == NULL){
fprintf(stderr, "Can't allocate line_buf[%d]\n", i);
exit(1);
}
}
do { // user が 1になった時にフレームがスタートする
ins >> pix;
} while(pix.user == 0);
for (int y=0; y<height; y++){
for (int x=0; x<width; x++){
if (!(x==0 && y==0)) // 最初の入力はすでに入力されている
ins >> pix; // AXI4-Stream からの入力
for (int k=0; k<3; k++){
for (int m=0; m<2; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
pix_mat[0][2] = line_buf[0][x];
pix_mat[1][2] = line_buf[1][x];
int y_val = conv_rgb2y_soft(pix.data);
pix_mat[2][2] = y_val;
line_buf[0][x] = line_buf[1][x]; // 行の入れ替え
line_buf[1][x] = y_val;
lap_fil_val = laplacian_fil_soft( pix_mat[0][0], pix_mat[0][1], pix_mat[0][2],
pix_mat[1][0], pix_mat[1][1], pix_mat[1][2],
pix_mat[2][0], pix_mat[2][1], pix_mat[2][2]);
lap.data = (lap_fil_val<<16)+(lap_fil_val<<8)+lap_fil_val; // RGB同じ値を入れる
if (x<2 || y<2) // 最初の2行とその他の行の最初の2列は無効データなので0とする
lap.data = 0;
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
lap.user = 1;
else
lap.user = 0;
if (x == (width-1)) // 行の最後で TLAST をアサートする
lap.last = 1;
else
lap.last = 0;
outs << lap; // AXI4-Stream へ出力
}
}
for (i=0; i<2; i++)
free(line_buf[i]);
free(line_buf);
return 0;
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y_soft(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
// ラプラシアンフィルタ
// x0y0 x1y0 x2y0 -1 -1 -1
// x0y1 x1y1 x2y1 -1 8 -1
// x0y2 x1y2 x2y2 -1 -1 -1
int laplacian_fil_soft(int x0y0, int x1y0, int x2y0, int x0y1, int x1y1, int x2y1, int x0y2, int x1y2, int x2y2)
{
int y;
y = -x0y0 -x1y0 -x2y0 -x0y1 +8*x1y1 -x2y1 -x0y2 -x1y2 -x2y2;
if (y<0)
y = 0;
else if (y>255)
y = 255;
return(y);
}
set_property PACKAGE_PIN R18 [get_ports {sw[3]}] # BTN0
sw0 - 0 : NORMAL, 1 : DEBUG
sw1 - 0 : MOTOR OFF, 1 : MOTOR ON
sw2 - 0 : 左白線用のパラメータでガボール・フィルタさせた画像をディスプレイに表示する
sw2 - 1 : 右白線用のパラメータでガボール・フィルタさせた画像をディスプレイに表示する
BTN0 - 1 : sw2 で示された SVGA 画面分のガボール・フィルタの値を標準出力に出力する
// wl_tracking_gabor_bm.c
// 2022/12/18 : by marsee
// 2022/12/21 : bug fix. by marsee
// 2022/12/23 : Added sw2, BTN0.
#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include "sleep.h"
#include "xparameters.h"
#include "xil_io.h"
#include "xpwm.h"
#include "xmotor_monitor.h"
#define DIR_LEFT_NORMAL 1
#define DIR_LEFT_REVERSE 0
#define DIR_RIGHT_NORMAL 0
#define DIR_RIGHT_REVERSE 1
#define PIXEL_NUM_OF_BYTES 4
#define SVGA_HORIZONTAL_PIXELS 800
#define SVGA_VERTICAL_LINES 600
#define SVGA_ALL_DISP_ADDRESS (SVGA_HORIZONTAL_PIXELS * SVGA_VERTICAL_LINES * PIXEL_NUM_OF_BYTES)
#define GABOR_DETECT_LINE 590
#define GABOR_DETECT_LINE_ADDR (SVGA_HORIZONTAL_PIXELS * GABOR_DETECT_LINE * PIXEL_NUM_OF_BYTES)
#define GABOR_THRESHOLD 255
#define DIST_THRESHOLD 30
#define LEFT_GABOR_EDGE_OVERFLOW 0
#define RIGHT_GABOR_EDGE_OVERFLOW (SVGA_HORIZONTAL_PIXELS/2)
//#define DEBUG
//#define MOTOR_OFF
#define FRAME_BUFFER_ADDRESS 0x10000000
#define GABOR_LEFT 0
#define GABOR_RIGHT 1
void cam_i2c_init(volatile uint32_t *mt9d111_axi_iic) {
mt9d111_axi_iic[64] = 0x2; // reset tx fifo ,address is 0x100, i2c_control_reg
mt9d111_axi_iic[64] = 0x1; // enable i2c
}
void cam_i2x_write_sync(void) {
// unsigned c;
// c = *cam_i2c_rx_fifo;
// while ((c & 0x84) != 0x80)
// c = *cam_i2c_rx_fifo; // No Bus Busy and TX_FIFO_Empty = 1
usleep(1000);
}
void cam_i2c_write(volatile uint32_t *mt9d111_axi_iic, uint32_t device_addr, uint32_t write_addr, uint32_t write_data){
mt9d111_axi_iic[66] = 0x100 | (device_addr & 0xfe); // Slave IIC Write Address, address is 0x108, i2c_tx_fifo
mt9d111_axi_iic[66] = write_addr;
mt9d111_axi_iic[66] = (write_data >> 8)|0xff; // first data
mt9d111_axi_iic[66] = 0x200 | (write_data & 0xff); // second data
cam_i2x_write_sync();
}
// Gobor filter
//
void gabor_fil_on(){
volatile uint32_t *axis_switch_0, *axis_switch_1;
volatile uint32_t *gabor_filter_lh_0;
axis_switch_0 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_AXIS_SWITCH_0_BASEADDR;
axis_switch_1 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_AXIS_SWITCH_1_BASEADDR;
gabor_filter_lh_0 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_GABOR_FILTER_LH_2_0_S_AXI_CONTROL_BASEADDR;
// axis_switch_1, 1to2 ,Select M01_AXIS
// Refer to http://marsee101.blog19.fc2.com/blog-entry-3177.html
axis_switch_1[16] = 0x80000000; // 0x40 = 0x80000000; axi_switch_0 M00_AXIS disable
axis_switch_1[17] = 0x80000000; // 0x44 = 0x80000000; axi_switch_0 M01_AXIS disable
axis_switch_1[18] = 0x80000000; // 0x48 = 0x80000000; axi_switch_0 M02_AXIS disable
axis_switch_1[19] = 0x80000000; // 0x4C = 0x80000000; axi_switch_0 M03_AXIS disable
axis_switch_1[20] = 0; // 0x50 = 0; axi_switch_4 M00_AXIS enable. gabor fitler
axis_switch_1[0] = 0x2; // 0x0 = 2; Commit registers
// gabor filter AXIS Start
gabor_filter_lh_0[0] = 0x01; // Start bit set
gabor_filter_lh_0[0] = 0x80; // Auto Restart bit set
// axis_switch_0, 2to1, Select S01_AXIS
// Refer to http://marsee101.blog19.fc2.com/blog-entry-3177.html
axis_switch_0[16] = 0x4; // 0x40 = 0x4; gabor fitler
axis_switch_0[0] = 0x2; // 0x0 = 2; Commit registers
}
int search_gabor_edge(uint32_t start_addr, uint32_t number, uint32_t threshold){
volatile uint32_t *imgaddr;
int i;
imgaddr = (volatile uint32_t *)start_addr;
//printf("start_addr = %x\n", (int)start_addr);
for (i=0; i<number; i++){
uint32_t c=imgaddr[i] & 0xff;
//printf("%d, %d, %d\n",(int)number, i, (int)c);
if (c >= threshold){
break;
}
}
return(i);
}
// Motor
//
void motor_settings(XPwm *motorLp, XPwm *motorRp){
XPwm_DisableAutoRestart(motorLp);
while(!XPwm_IsIdle(motorLp)) ;
XPwm_Start(motorLp);
XPwm_EnableAutoRestart(motorLp);
XPwm_DisableAutoRestart(motorRp);
while(!XPwm_IsIdle(motorRp)) ;
XPwm_Start(motorRp);
XPwm_EnableAutoRestart(motorRp);
}
void Stopped_Zybot(XPwm *motorLp, XPwm *motorRp){
XPwm_Set_sw_late_V(motorLp, 0);
XPwm_Set_sw_late_V(motorRp, 0);
}
void motor_initialize(XPwm *motorL, XPwm *motorR, XMotor_monitor *mmL, XMotor_monitor *mmR){
XPwm *motorLp, *motorRp;
XMotor_monitor *mmLp, *mmRp;
motorLp = motorL;
motorRp = motorR;
mmLp = mmL;
mmRp = mmR;
// Initialization of motor
if (XPwm_Initialize(motorLp, 0) != XST_SUCCESS){
fprintf(stderr,"pwm_0 (Left) open error\n");
exit(-1);
}
if (XPwm_Initialize(motorRp, 1) != XST_SUCCESS){
fprintf(stderr,"pwm_1 (Right) open error\n");
exit(-1);
}
// Initialization of motor monitor
if (XMotor_monitor_Initialize(mmLp, 0) != XST_SUCCESS){
fprintf(stderr,"motor_monitor_0 (Left) open error\n");
exit(-1);
}
if (XMotor_monitor_Initialize(mmRp, 1) != XST_SUCCESS){
fprintf(stderr,"motor_monitor_1 (Right) open error\n");
exit(-1);
}
// The Motors is rotated in the forward direction.
XPwm_Set_sw_late_V(motorLp, 0);
XPwm_Set_dir_V(motorLp, 1);
XPwm_Set_sw_late_V(motorRp, 0);
XPwm_Set_dir_V(motorRp, 0);
motor_settings(motorLp, motorRp);
}
int check_debug(volatile uint32_t *axi_gpio_0){
uint32_t val = axi_gpio_0[0];
return(val & 1);
}
int check_motor_on(volatile uint32_t *axi_gpio_0){
uint32_t val = axi_gpio_0[0];
val = val>>1;
return(val & 1);
}
int check_gabor_LR(volatile uint32_t *axi_gpio_0){
uint32_t val = axi_gpio_0[0];
val = val>>2;
return(val & 1);
}
int check_gabor_display(volatile uint32_t *axi_gpio_0){
uint32_t val = axi_gpio_0[0];
val = val>>3;
return(val & 1);
}
int gabor_data_disp(uint32_t fb_addr, uint32_t lr){
volatile uint32_t *fb_addrp = (volatile uint32_t *)fb_addr;
if(lr == GABOR_LEFT)
printf("----- gabor left data -----\n");
else
printf("----- gabor right data -----\n");
for(int i=0; i<SVGA_HORIZONTAL_PIXELS*SVGA_VERTICAL_LINES; i++){
if((i%40) == 0)
printf("\n");
printf("%02x", (int)(fb_addrp[i] & 0xff));
}
printf("\n");
return(0);
}
int main(){
volatile uint32_t *dmaw4gabor_0;
XPwm motorL, motorR;
XMotor_monitor mmL, mmR;
int left_wl_edge, right_wl_edge;
volatile uint32_t *frame_buffer;
// mt9d111_inf_axis_0, axi_iic_0, bitmap_disp_cntrler_axi_master_0
volatile uint32_t *bmdc0_axi_lites;
volatile uint32_t *mt9d111_axi_lites;
volatile uint32_t *mt9d111_i2c_axi_lites;
volatile uint32_t *axi_gpio_0;
// axi_gpio_0 : sw0 - 0 : NORMAL, 1 : DEBUG
// sw1 - 0 : MOTOR OFF, 1 : MOTOR ON
// sw2 - 0 : Display the Gabor-filtered image for the left white line on the display
// sw2 - 1 : Display the Gabor-filtered image for the right white line on the display
// BTN0 - 1 : Display the value of the Gabor filter for one screen. sw2 designates right white line data or left white line data.
axi_gpio_0 = (volatile uint32_t *)XPAR_AXI_GPIO_0_BASEADDR;
// Motor Initialize
motor_initialize(&motorL, &motorR, &mmL, &mmR);
// DMAW4Gabor Initialize
dmaw4gabor_0 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_DMAW4GABOR_0_S_AXI_AXILITES_BASEADDR;
frame_buffer = (volatile uint32_t *)FRAME_BUFFER_ADDRESS;
// DMA4Gabor frame_buffer setting
dmaw4gabor_0[6] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // Data signal of frame_buffer0
dmaw4gabor_0[8] = (volatile uint32_t)FRAME_BUFFER_ADDRESS + (volatile uint32_t)SVGA_ALL_DISP_ADDRESS; // Data signal of frame_buffer1
// Camera, display controller
bmdc0_axi_lites = (volatile uint32_t *)XPAR_BITMAP_DISP_CNTRLER_AXI_MASTER_0_BASEADDR;
mt9d111_axi_lites = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_MT9D111_INF_AXIS_0_BASEADDR;
mt9d111_i2c_axi_lites = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_AXI_IIC_0_BASEADDR;
// IP start
dmaw4gabor_0[0] = (volatile uint32_t)0x81; // start, auto restart
gabor_fil_on();
bmdc0_axi_lites[0] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // Bitmap Display Controller 0 start
mt9d111_axi_lites[0] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // Camera Interface start (Address is dummy)
// CMOS Camera initialize, MT9D111
cam_i2c_init(mt9d111_i2c_axi_lites);
cam_i2c_write(mt9d111_i2c_axi_lites, 0xba, 0xf0, 0x1); // Changed regster map to IFP page 1
cam_i2c_write(mt9d111_i2c_axi_lites, 0xba, 0x97, 0x20); // RGB Mode, RGB565
mt9d111_axi_lites[1] = 0; // One_shot_mode is disabled
// main loop
if(check_debug(axi_gpio_0))
printf("White line Tracking start. \n");
while(1){
// Gabor filter for left white line
left_wl_edge = SVGA_HORIZONTAL_PIXELS/2 - search_gabor_edge(
FRAME_BUFFER_ADDRESS+GABOR_DETECT_LINE_ADDR, SVGA_HORIZONTAL_PIXELS/2, GABOR_THRESHOLD);
// Gabor filter for right white line
right_wl_edge = search_gabor_edge(
FRAME_BUFFER_ADDRESS+SVGA_ALL_DISP_ADDRESS+GABOR_DETECT_LINE_ADDR+(SVGA_HORIZONTAL_PIXELS/2)*PIXEL_NUM_OF_BYTES,
SVGA_HORIZONTAL_PIXELS/2, GABOR_THRESHOLD);
if(check_debug(axi_gpio_0))
printf("left_wl_edge = %d, right_wl_edge = %d\n", left_wl_edge, right_wl_edge);
if (left_wl_edge == LEFT_GABOR_EDGE_OVERFLOW){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 15);
XPwm_Set_sw_late_V(&motorR, 45);
}
if(check_debug(axi_gpio_0))
printf("Left gabor edge is overflow\n");
} else if (right_wl_edge == RIGHT_GABOR_EDGE_OVERFLOW){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 45);
XPwm_Set_sw_late_V(&motorR, 15);
}
if(check_debug(axi_gpio_0))
printf("Right gabar edge is overflow\n");
} else if ((right_wl_edge - left_wl_edge) > DIST_THRESHOLD){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 35);
XPwm_Set_sw_late_V(&motorR, 25);
}
if(check_debug(axi_gpio_0))
printf("Right turn\n");
} else if ((right_wl_edge - left_wl_edge) < -DIST_THRESHOLD){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 25);
XPwm_Set_sw_late_V(&motorR, 35);
}
if(check_debug(axi_gpio_0))
printf("Left turn\n");
} else if (abs(right_wl_edge - left_wl_edge) <= DIST_THRESHOLD){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 30);
XPwm_Set_sw_late_V(&motorR, 30);
}
if(check_debug(axi_gpio_0))
printf("Go straight\n");
}
if(check_gabor_LR(axi_gpio_0)){
bmdc0_axi_lites[0] = (volatile uint32_t)FRAME_BUFFER_ADDRESS + (volatile uint32_t)SVGA_ALL_DISP_ADDRESS; // right
} else {
bmdc0_axi_lites[0] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // left
}
if(check_gabor_display(axi_gpio_0)){
if(check_gabor_LR(axi_gpio_0)){
gabor_data_disp((uint32_t)FRAME_BUFFER_ADDRESS+(uint32_t)SVGA_ALL_DISP_ADDRESS, (uint32_t)GABOR_RIGHT);
} else {
gabor_data_disp((uint32_t)FRAME_BUFFER_ADDRESS, (uint32_t)GABOR_LEFT);
}
}
if(!check_motor_on(axi_gpio_0)){ // sw1 off. Motor Stopped
XPwm_Set_sw_late_V(&motorL, 0);
XPwm_Set_sw_late_V(&motorR, 0);
}
}
}
// wl_tracking_gabor_bm.c
// 2022/12/18 : by marsee
// 2022/12/21 : bug fix. by marsee
#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include "sleep.h"
#include "xparameters.h"
#include "xil_io.h"
#include "xpwm.h"
#include "xmotor_monitor.h"
#define DIR_LEFT_NORMAL 1
#define DIR_LEFT_REVERSE 0
#define DIR_RIGHT_NORMAL 0
#define DIR_RIGHT_REVERSE 1
#define PIXEL_NUM_OF_BYTES 4
#define SVGA_HORIZONTAL_PIXELS 800
#define SVGA_VERTICAL_LINES 600
#define SVGA_ALL_DISP_ADDRESS (SVGA_HORIZONTAL_PIXELS * SVGA_VERTICAL_LINES * PIXEL_NUM_OF_BYTES)
#define GABOR_DETECT_LINE 590
#define GABOR_DETECT_LINE_ADDR (SVGA_HORIZONTAL_PIXELS * GABOR_DETECT_LINE * PIXEL_NUM_OF_BYTES)
#define GABOR_THRESHOLD 5
#define DIST_THRESHOLD 30
#define LEFT_GABOR_EDGE_OVERFLOW 0
#define RIGHT_GABOR_EDGE_OVERFLOW (SVGA_HORIZONTAL_PIXELS/2)
//#define DEBUG
//#define MOTOR_OFF
#define FRAME_BUFFER_ADDRESS 0x10000000
void cam_i2c_init(volatile uint32_t *mt9d111_axi_iic) {
mt9d111_axi_iic[64] = 0x2; // reset tx fifo ,address is 0x100, i2c_control_reg
mt9d111_axi_iic[64] = 0x1; // enable i2c
}
void cam_i2x_write_sync(void) {
// unsigned c;
// c = *cam_i2c_rx_fifo;
// while ((c & 0x84) != 0x80)
// c = *cam_i2c_rx_fifo; // No Bus Busy and TX_FIFO_Empty = 1
usleep(1000);
}
void cam_i2c_write(volatile uint32_t *mt9d111_axi_iic, uint32_t device_addr, uint32_t write_addr, uint32_t write_data){
mt9d111_axi_iic[66] = 0x100 | (device_addr & 0xfe); // Slave IIC Write Address, address is 0x108, i2c_tx_fifo
mt9d111_axi_iic[66] = write_addr;
mt9d111_axi_iic[66] = (write_data >> 8)|0xff; // first data
mt9d111_axi_iic[66] = 0x200 | (write_data & 0xff); // second data
cam_i2x_write_sync();
}
// Gobor filter
//
void gabor_fil_on(){
volatile uint32_t *axis_switch_0, *axis_switch_1;
volatile uint32_t *gabor_filter_lh_0;
axis_switch_0 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_AXIS_SWITCH_0_BASEADDR;
axis_switch_1 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_AXIS_SWITCH_1_BASEADDR;
gabor_filter_lh_0 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_GABOR_FILTER_LH_2_0_S_AXI_CONTROL_BASEADDR;
// axis_switch_1, 1to2 ,Select M01_AXIS
// Refer to http://marsee101.blog19.fc2.com/blog-entry-3177.html
axis_switch_1[16] = 0x80000000; // 0x40 = 0x80000000; axi_switch_0 M00_AXIS disable
axis_switch_1[17] = 0x80000000; // 0x44 = 0x80000000; axi_switch_0 M01_AXIS disable
axis_switch_1[18] = 0x80000000; // 0x48 = 0x80000000; axi_switch_0 M02_AXIS disable
axis_switch_1[19] = 0x80000000; // 0x4C = 0x80000000; axi_switch_0 M03_AXIS disable
axis_switch_1[20] = 0; // 0x50 = 0; axi_switch_4 M00_AXIS enable. gabor fitler
axis_switch_1[0] = 0x2; // 0x0 = 2; Commit registers
// gabor filter AXIS Start
gabor_filter_lh_0[0] = 0x01; // Start bit set
gabor_filter_lh_0[0] = 0x80; // Auto Restart bit set
// axis_switch_0, 2to1, Select S01_AXIS
// Refer to http://marsee101.blog19.fc2.com/blog-entry-3177.html
axis_switch_0[16] = 0x4; // 0x40 = 0x4; gabor fitler
axis_switch_0[0] = 0x2; // 0x0 = 2; Commit registers
}
int search_gabor_edge(unsigned int start_addr, unsigned int number, int threshold){
volatile int *imgaddr;
int i;
imgaddr = (volatile int *)start_addr;
for (i=0; i<number; i++){
int c=imgaddr[i] & 0xff;
//printf("%d\n",c);
if (c >= threshold){
break;
}
}
return(i);
}
// Motor
//
void motor_settings(XPwm *motorLp, XPwm *motorRp){
XPwm_DisableAutoRestart(motorLp);
while(!XPwm_IsIdle(motorLp)) ;
XPwm_Start(motorLp);
XPwm_EnableAutoRestart(motorLp);
XPwm_DisableAutoRestart(motorRp);
while(!XPwm_IsIdle(motorRp)) ;
XPwm_Start(motorRp);
XPwm_EnableAutoRestart(motorRp);
}
void Stopped_Zybot(XPwm *motorLp, XPwm *motorRp){
XPwm_Set_sw_late_V(motorLp, 0);
XPwm_Set_sw_late_V(motorRp, 0);
}
void motor_initialize(XPwm *motorL, XPwm *motorR, XMotor_monitor *mmL, XMotor_monitor *mmR){
XPwm *motorLp, *motorRp;
XMotor_monitor *mmLp, *mmRp;
motorLp = motorL;
motorRp = motorR;
mmLp = mmL;
mmRp = mmR;
// Initialization of motor
if (XPwm_Initialize(motorLp, 0) != XST_SUCCESS){
fprintf(stderr,"pwm_0 (Left) open error\n");
exit(-1);
}
if (XPwm_Initialize(motorRp, 1) != XST_SUCCESS){
fprintf(stderr,"pwm_1 (Right) open error\n");
exit(-1);
}
// Initialization of motor monitor
if (XMotor_monitor_Initialize(mmLp, 0) != XST_SUCCESS){
fprintf(stderr,"motor_monitor_0 (Left) open error\n");
exit(-1);
}
if (XMotor_monitor_Initialize(mmRp, 1) != XST_SUCCESS){
fprintf(stderr,"motor_monitor_1 (Right) open error\n");
exit(-1);
}
// The Motors is rotated in the forward direction.
XPwm_Set_sw_late_V(motorLp, 0);
XPwm_Set_dir_V(motorLp, 1);
XPwm_Set_sw_late_V(motorRp, 0);
XPwm_Set_dir_V(motorRp, 0);
motor_settings(motorLp, motorRp);
}
int check_debug(volatile uint32_t *axi_gpio_0){
uint32_t val = axi_gpio_0[0];
return(val & 1);
}
int check_motor_on(volatile uint32_t *axi_gpio_0){
uint32_t val = axi_gpio_0[0];
val >>= 1;
return(val & 1);
}
int main(){
volatile uint32_t *dmaw4gabor_0;
XPwm motorL, motorR;
XMotor_monitor mmL, mmR;
int left_wl_edge, right_wl_edge;
volatile uint32_t *frame_buffer;
// mt9d111_inf_axis_0, axi_iic_0, bitmap_disp_cntrler_axi_master_0
volatile uint32_t *bmdc0_axi_lites;
volatile uint32_t *mt9d111_axi_lites;
volatile uint32_t *mt9d111_i2c_axi_lites;
volatile uint32_t *axi_gpio_0;
// axi_gpio_0 : sw0 - 0 : NORMAL, 1 : DEBUG
// sw1 - 0 : MOTOR OFF, 1 : MOTOR ON
axi_gpio_0 = (volatile uint32_t *)XPAR_AXI_GPIO_0_BASEADDR;
// Motor Initialize
motor_initialize(&motorL, &motorR, &mmL, &mmR);
// DMAW4Gabor Initialize
dmaw4gabor_0 = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_DMAW4GABOR_0_S_AXI_AXILITES_BASEADDR;
frame_buffer = (volatile uint32_t *)FRAME_BUFFER_ADDRESS;
// DMA4Gabor frame_buffer setting
dmaw4gabor_0[6] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // Data signal of frame_buffer0
dmaw4gabor_0[8] = (volatile uint32_t)FRAME_BUFFER_ADDRESS + (volatile uint32_t)SVGA_ALL_DISP_ADDRESS; // Data signal of frame_buffer1
// Camera, display controller
bmdc0_axi_lites = (volatile uint32_t *)XPAR_BITMAP_DISP_CNTRLER_AXI_MASTER_0_BASEADDR;
mt9d111_axi_lites = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_MT9D111_INF_AXIS_0_BASEADDR;
mt9d111_i2c_axi_lites = (volatile uint32_t *)XPAR_CAMERA_INTERFACE_AXI_IIC_0_BASEADDR;
// IP start
dmaw4gabor_0[0] = (volatile uint32_t)0x81; // start, auto restart
gabor_fil_on();
bmdc0_axi_lites[0] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // Bitmap Display Controller 0 start
mt9d111_axi_lites[0] = (volatile uint32_t)FRAME_BUFFER_ADDRESS; // Camera Interface start (Address is dummy)
// CMOS Camera initialize, MT9D111
cam_i2c_init(mt9d111_i2c_axi_lites);
cam_i2c_write(mt9d111_i2c_axi_lites, 0xba, 0xf0, 0x1); // Changed regster map to IFP page 1
cam_i2c_write(mt9d111_i2c_axi_lites, 0xba, 0x97, 0x20); // RGB Mode, RGB565
mt9d111_axi_lites[1] = 0; // One_shot_mode is disabled
// main loop
if(check_debug(axi_gpio_0))
printf("White line Tracking start. \n");
while(1){
// Gabor filter for left white line
left_wl_edge = SVGA_HORIZONTAL_PIXELS/2 - search_gabor_edge(
(unsigned int)frame_buffer+GABOR_DETECT_LINE_ADDR, SVGA_HORIZONTAL_PIXELS/2, GABOR_THRESHOLD);
// Gabor filter for right white line
right_wl_edge = search_gabor_edge(
(unsigned int)frame_buffer+SVGA_ALL_DISP_ADDRESS+GABOR_DETECT_LINE_ADDR+(SVGA_HORIZONTAL_PIXELS/2)*PIXEL_NUM_OF_BYTES,
SVGA_HORIZONTAL_PIXELS/2, GABOR_THRESHOLD);
if(check_debug(axi_gpio_0))
printf("left_wl_edge = %d, right_wl_edge = %d\n", left_wl_edge, right_wl_edge);
if (left_wl_edge == LEFT_GABOR_EDGE_OVERFLOW){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 15);
XPwm_Set_sw_late_V(&motorR, 45);
}
if(check_debug(axi_gpio_0))
printf("Left gabor edge is overflow\n");
} else if (right_wl_edge == RIGHT_GABOR_EDGE_OVERFLOW){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 45);
XPwm_Set_sw_late_V(&motorR, 15);
}
if(check_debug(axi_gpio_0))
printf("Right gabar edge is overflow\n");
} else if ((right_wl_edge - left_wl_edge) > DIST_THRESHOLD){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 35);
XPwm_Set_sw_late_V(&motorR, 25);
}
if(check_debug(axi_gpio_0))
printf("Right turn\n");
} else if ((right_wl_edge - left_wl_edge) < -DIST_THRESHOLD){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 25);
XPwm_Set_sw_late_V(&motorR, 35);
}
if(check_debug(axi_gpio_0))
printf("Left turn\n");
} else if (abs(right_wl_edge - left_wl_edge) <= DIST_THRESHOLD){
if(check_motor_on(axi_gpio_0)){
XPwm_Set_sw_late_V(&motorL, 30);
XPwm_Set_sw_late_V(&motorR, 30);
}
if(check_debug(axi_gpio_0))
printf("Go straight\n");
}
}
}
// Gabor_filter_lh_2.h
// 2016/07/24
// 2016/07/25 : 右白線検出用のGabor Filterの重みを追加
// 2016/07/27 : 右白線検出用配列と左白線検出用配列を統合
// 2016/08/29 : 1回目左白線検出、2回目右白線検出のモードを追加
//
#ifndef __Gabor_filter_lh_H__
#define __Gabor_filter_lh_H__
#define HORIZONTAL_PIXEL_WIDTH 800
#define VERTICAL_PIXEL_WIDTH 600
//#define HORIZONTAL_PIXEL_WIDTH 640 // for Simulation
//#define VERTICAL_PIXEL_WIDTH 480
//#define HORIZONTAL_PIXEL_WIDTH 64
//#define VERTICAL_PIXEL_WIDTH 48
#define ALL_PIXEL_VALUE (HORIZONTAL_PIXEL_WIDTH*VERTICAL_PIXEL_WIDTH)
#define ARRAY_SIZE 9
#define RIGHT_WEIGHT 1
#define LEFT_WEIGHT 0
#define L_R_WEIGHT 2
const int gabor_weight[2][ARRAY_SIZE][ARRAY_SIZE] = { // 左白線検出用＋右白線検出用
{
{0,-3,-10,-16,-7,7,10,5,1},
{-3,-15,-27,-11,32,50,29,5,-2},
{-11,-24,-5,73,135,95,16,-17,-10},
{-11,4,85,187,160,14,-72,-52,-14},
{4,51,135,137,-18,-159,-136,-45,-2},
{16,50,59,-39,-179,-185,-73,3,13},
{10,12,-25,-104,-131,-60,15,27,11},
{1,-7,-31,-48,-24,18,29,14,3},
{-1,-5,-9,-4,10,16,10,2,-1}
},
{
{1,5,7,1,-12,-17,-10,-3,0},
{1,11,33,45,21,-16,-27,-15,-4},
{-8,-5,35,107,131,69,-2,-21,-11},
{-17,-47,-51,40,169,187,93,13,-7},
{-8,-54,-134,-147,-18,123,130,58,11},
{9,-6,-82,-185,-187,-65,36,44,17},
{11,24,12,-55,-125,-112,-43,1,7},
{3,14,30,23,-13,-41,-33,-12,-1},
{0,2,9,17,14,1,-6,-5,-2}
}
};
const float gabor_fweight[2][ARRAY_SIZE][ARRAY_SIZE] = { // 左白線検出用＋右白線検出用(float)
{
{0.001282,-0.009914,-0.04062,-0.060586,-0.027574,0.026072,0.038427,0.018191,0.003056},
{-0.012155,-0.057448,-0.104645,-0.042953,0.123263,0.197238,0.114451,0.020448,-0.007239},
{-0.042252,-0.093065,-0.018911,0.285374,0.525746,0.372687,0.060734,-0.064748,-0.040465},
{-0.042261,0.015673,0.332798,0.728763,0.625046,0.053591,-0.283076,-0.203293,-0.05608},
{0.017342,0.198305,0.52554,0.535526,-0.069756,-0.622839,-0.531089,-0.177366,-0.006367},
{0.060866,0.19708,0.231032,-0.154219,-0.699885,-0.721808,-0.286707,0.013004,0.049249},
{0.038379,0.04877,-0.098477,-0.404993,-0.510165,-0.233566,0.057894,0.104366,0.041887},
{0.0047,-0.0278,-0.121277,-0.187262,-0.093276,0.070512,0.113857,0.055799,0.009976},
{-0.003798,-0.01885,-0.035607,-0.01709,0.037692,0.064268,0.038606,0.007536,-0.002133}
},
{
{0.005562,0.018882,0.028293,0.004499,-0.044995,-0.064838,-0.039469,-0.009822,0.000815},
{0.002294,0.04108,0.127023,0.175094,0.083025,-0.063755,-0.106402,-0.057798,-0.01406},
{-0.031269,-0.021096,0.135641,0.417286,0.512467,0.269946,-0.008354,-0.082091,-0.041357},
{-0.066348,-0.184919,-0.197802,0.15614,0.65976,0.728616,0.361674,0.052074,-0.027152},
{-0.031146,-0.211178,-0.523777,-0.573856,-0.069756,0.480311,0.506451,0.225223,0.041031},
{0.035552,-0.023892,-0.320104,-0.723563,-0.728735,-0.253689,0.1391,0.170625,0.067723},
{0.04216,0.094939,0.047511,-0.216623,-0.488075,-0.437898,-0.168739,0.003336,0.027009},
{0.012112,0.056596,0.115239,0.090332,-0.05076,-0.158403,-0.127847,-0.046375,-0.004918},
{-0.00168,0.007437,0.036985,0.067021,0.053689,0.004977,-0.02365,-0.018248,-0.005928}
}
};
#endif
// Gabor_fiter_lh_2.cpp
// 2016/07/23 by marsee
// 2016/07/25 : 右白線検出用のGabor Filterを追加して、右左の白線を指定するRorL 引数を追加
// 2016/07/27 : 右白線検出用配列と左白線検出用配列を統合
// 2016/08/29 : 1回目左白線検出、2回目右白線検出のモードを追加
// 2022/12/14 : Vitis HLS用に書き直した
//
#include <stdio.h>
#include <string.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <ap_axi_sdata.h>
#include "Gabor_filter_lh_2.h"
int conv_rgb2y(int rgb);
int Gabor_filter_lh_2(hls::stream<ap_axis<32,1,1,1> >& ins,
hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL){
#pragma HLS INTERFACE mode=ap_none port=RorL
#pragma HLS INTERFACE mode=axis register_mode=both port=ins register
#pragma HLS INTERFACE mode=axis register_mode=both port=outs register
#pragma HLS INTERFACE mode=s_axilite port=return
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> gabor;
ap_int<32> line_buf[ARRAY_SIZE-1][1920]; // Up to HD resolution
#pragma HLS array_partition variable=line_buf block factor=2 dim=1
ap_int<32> pix_mat[ARRAY_SIZE][ARRAY_SIZE];
int gray_pix, val, i, j, x, y, k;
LOOP_WAIT_USER: do {
#pragma HLS LOOP_TRIPCOUNT avg=1 max=1 min=1
// user が 1になった時にフレームがスタートする
ins >> pix;
} while(pix.user == 0);
LOOP_LR: for (k=0; k<2; k++){
LOOP_Y: for (y=0; y<VERTICAL_PIXEL_WIDTH; y++){
LOOP_X: for (x=0; x<HORIZONTAL_PIXEL_WIDTH; x++){
#pragma HLS PIPELINE II=1
if (!(k==0 && x==0 && y==0)) // 最初の入力はすでに入力されている
ins >> pix; // AXI4-Stream からの入力
LOOP_PIX_MAT_K: for(int k=0; k<ARRAY_SIZE; k++){
LOOP_PIX_MAT_M: for(int m=0; m<ARRAY_SIZE-1; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
for (i=0; i<ARRAY_SIZE-1; i++){
pix_mat[i][ARRAY_SIZE-1] = line_buf[i][x];
}
ap_int<32> y_val = conv_rgb2y(pix.data);
pix_mat[ARRAY_SIZE-1][ARRAY_SIZE-1] = y_val;
for (i=0; i<ARRAY_SIZE-2; i++){ // 行の入れ替え
line_buf[i][x] = line_buf[i+1][x];
}
line_buf[ARRAY_SIZE-2][x] = y_val;
// 使用する配列を決定する
int ano;
switch (RorL){
case LEFT_WEIGHT :
ano = LEFT_WEIGHT;
break;
case RIGHT_WEIGHT :
ano = RIGHT_WEIGHT;
break;
case L_R_WEIGHT :
if (k == 0)
ano = LEFT_WEIGHT;
else
ano = RIGHT_WEIGHT;
break;
default :
ano = L_R_WEIGHT;
break;
}
// Gabor filter の演算
for (j=0, val=0; j<ARRAY_SIZE; j++){
for (i=0; i<ARRAY_SIZE; i++){
val += gabor_weight[ano][j][i] * pix_mat[j][i];
}
}
val = val/256; // 256倍してあるので、1/256して戻す
if (val<0)
//val = -val; // 絶対値
val = 0; // マイナスの値を0に丸める
else if (val>255)
val = 255;
// Gabor filter・データの書き込み
gabor.data = (val<<16)+(val<<8)+val;
// 最初のARRAY_SIZE-1行とその他の行の最初のARRAY_SIZE-1列は無効データなので0とする
if (x<(ARRAY_SIZE-1) || y<(ARRAY_SIZE-1))
gabor.data = 0;
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
gabor.user = 1;
else
gabor.user = 0;
if (x == (HORIZONTAL_PIXEL_WIDTH-1)) // 行の最後で TLAST をアサートする
gabor.last = 1;
else
gabor.last = 0;
outs << gabor; // AXI4-Stream へ出力
}
}
}
return(0);
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
// Gabor_filter_lh_2_tb.cpp
// 2016/07/24 by marsee
// 2016/07/25 : 右白線検出用のGabor Filterを追加して、右左の白線を指定するRorL 引数を追加
// 2022/12/14 : Vitis HLS用に書き直した
//
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ap_int.h>
#include <hls_stream.h>
#include <iostream>
#include <fstream>
#include <math.h>
#include <ap_axi_sdata.h>
#include "Gabor_filter_lh_2.h"
#include "bmp_header.h"
int Gabor_filter_lh_2(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL);
int conv_rgb2y_soft(int rgb);
int Gabor_filter_lh_2_soft(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL);
#define CLOCK_PERIOD 10
#define RIGHT_OR_LEFT L_R_WEIGHT
#define BMP_FILE_NAME "road_1.bmp"
int main()
{
using namespace std;
hls::stream<ap_axis<32,1,1,1> > ins;
hls::stream<ap_axis<32,1,1,1> > ins_soft;
hls::stream<ap_axis<32,1,1,1> > outs;
hls::stream<ap_axis<32,1,1,1> > outs_soft;
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> vals;
ap_axis<32,1,1,1> vals_soft;
int m_seq = 1; // M系列の値
int i, k;
int xor_shift;
BITMAPFILEHEADER bmpfhr; // BMPファイルのファイルヘッダ(for Read)
BITMAPINFOHEADER bmpihr; // BMPファイルのINFOヘッダ(for Read)
FILE *fbmpr, *fbmpw, *fbmpwf;
int *rd_bmp, *hw_gabor, *sw_gabor;
int blue, green, red;
ap_uint<2> r_l;
char fhname[100];
char fsname[100];
if ((fbmpr = fopen(BMP_FILE_NAME, "rb")) == NULL){ // test.bmp をオープン
fprintf(stderr, "Can't open test.bmp by binary read mode\n");
exit(1);
}
// bmpヘッダの読み出し
fread(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpr);
fread(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpr);
fread(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpr);
fread(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpr);
printf("bmpihr.biWidth = %d, bmpihr.biHeight = %d\n",(int)bmpihr.biWidth, (int)bmpihr.biHeight);
// ピクセルを入れるメモリをアロケートする
if ((rd_bmp =(int *)malloc(sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate rd_bmp memory\n");
exit(1);
}
if ((hw_gabor =(int *)malloc(2 * sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate hw_gabor memory\n");
exit(1);
}
if ((sw_gabor =(int *)malloc(2 * sizeof(int) * (bmpihr.biWidth * bmpihr.biHeight))) == NULL){
fprintf(stderr, "Can't allocate hw_gabor memory\n");
exit(1);
}
// rd_bmp にBMPのピクセルを代入。その際に、行を逆転する必要がある
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = fgetc(fbmpr);
green = fgetc(fbmpr);
red = fgetc(fbmpr);
rd_bmp[((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] = (blue & 0xff) | ((green & 0xff)<<8) | ((red & 0xff)<<16);
}
}
fclose(fbmpr);
// ins に入力データを用意する
for(int i=0; i<5; i++){ // dummy data
pix.user = 0;
pix.data = i;
ins << pix;
}
// 2画面分を入力する
for(int k=0; k<2; k++){
for(int j=0; j < bmpihr.biHeight; j++){
for(i=0; i < bmpihr.biWidth; i++){
pix.data = (ap_int<32>)rd_bmp[(j*bmpihr.biWidth)+i];
if (j==0 && i==0) // 最初のデータの時に TUSER を 1 にする
pix.user = 1;
else
pix.user = 0;
if (i == bmpihr.biWidth-1) // 行の最後でTLASTをアサートする
pix.last = 1;
else
pix.last = 0;
ins << pix;
ins_soft << pix;
}
}
}
r_l = (ap_uint<2>)RIGHT_OR_LEFT;
Gabor_filter_lh_2(ins, outs, r_l);
Gabor_filter_lh_2_soft(ins_soft, outs_soft, r_l);
// ハードウェアとソフトウェアのラプラシアン・フィルタの値のチェック
cout << endl;
cout << "outs" << endl;
for(k=0; k<2; k++){
for(int j=0; j < bmpihr.biHeight; j++){
for(i=0; i < bmpihr.biWidth; i++){
outs >> vals;
outs_soft >> vals_soft;
ap_int<32> val = vals.data;
ap_int<32> val_soft = vals_soft.data;
hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+(j*bmpihr.biWidth)+i] = (int)val;
sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+(j*bmpihr.biWidth)+i] = (int)val_soft;
//printf("k=%d, j=%d, i=%d\n",k,j,i);
if ((double)pow((double)(val&0xff)-(val_soft&0xff),(double)2) > 4){ // 2乗誤差が4よりも大きい
printf("ERROR HW and SW results mismatch i = %ld, j = %ld, HW = %08x, SW = %08x\n", i, j, (int)val, (int)val_soft);
//return(1);
}
//if (vals.last)
//cout << "AXI-Stream is end" << endl;
}
}
}
cout << "Success HW and SW results match" << endl;
cout << endl;
// ハードウェアのラプラシアンフィルタの結果を temp_gabor0.bmp, temp_gabor1.bmp へ出力する
for(k=0; k<2; k++){
if(k == 0){
if ((fbmpw=fopen("temp_gabor0.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor0.bmp by binary write mode\n");
exit(1);
}
} else {
if ((fbmpw=fopen("temp_gabor1.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor1.bmp by binary write mode\n");
exit(1);
}
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpw);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpw);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpw);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (hw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpw);
fputc(green, fbmpw);
fputc(red, fbmpw);
}
}
fclose(fbmpw);
}
// ソフトウェアのラプラシアンフィルタの結果を temp_gabor_float0.bmp, temp_gabor_float1.bmp へ出力する
for(k=0; k<2; k++){
if (k == 0){
if ((fbmpwf=fopen("temp_gabor_float0.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor_float0.bmp by binary write mode\n");
exit(1);
}
} else {
if ((fbmpwf=fopen("temp_gabor_float1.bmp", "wb")) == NULL){
fprintf(stderr, "Can't open temp_gabor_float1.bmp by binary write mode\n");
exit(1);
}
}
// BMPファイルヘッダの書き込み
fwrite(&bmpfhr.bfType, sizeof(uint16_t), 1, fbmpwf);
fwrite(&bmpfhr.bfSize, sizeof(uint32_t), 1, fbmpwf);
fwrite(&bmpfhr.bfReserved1, sizeof(uint16_t), 1, fbmpwf);
fwrite(&bmpfhr.bfReserved2, sizeof(uint16_t), 1, fbmpwf);
fwrite(&bmpfhr.bfOffBits, sizeof(uint32_t), 1, fbmpwf);
fwrite(&bmpihr, sizeof(BITMAPINFOHEADER), 1, fbmpwf);
// RGB データの書き込み、逆順にする
for (int y=0; y<bmpihr.biHeight; y++){
for (int x=0; x<bmpihr.biWidth; x++){
blue = sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] & 0xff;
green = (sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x] >> 8) & 0xff;
red = (sw_gabor[bmpihr.biWidth*bmpihr.biHeight*k+((bmpihr.biHeight-1)-y)*bmpihr.biWidth+x]>>16) & 0xff;
fputc(blue, fbmpwf);
fputc(green, fbmpwf);
fputc(red, fbmpwf);
}
}
fclose(fbmpwf);
}
free(rd_bmp);
free(hw_gabor);
return 0;
}
int Gabor_filter_lh_2_soft(hls::stream<ap_axis<32,1,1,1> >& ins, hls::stream<ap_axis<32,1,1,1> >& outs, ap_uint<2> & RorL){
ap_axis<32,1,1,1> pix;
ap_axis<32,1,1,1> gabor;
ap_int<32> line_buf[ARRAY_SIZE-1][1920]; // Up to HD resolution
ap_int<32> pix_mat[ARRAY_SIZE][ARRAY_SIZE];
int gray_pix, val, i, j, x, y, k;
do {
// user が 1になった時にフレームがスタートする
ins >> pix;
} while(pix.user == 0);
for (k=0; k<2; k++){
for (y=0; y<VERTICAL_PIXEL_WIDTH; y++){
for (x=0; x<HORIZONTAL_PIXEL_WIDTH; x++){
if (!(k==0 && x==0 && y==0)) // 最初の入力はすでに入力されている
ins >> pix; // AXI4-Stream からの入力
for(int k=0; k<ARRAY_SIZE; k++){
for(int m=0; m<ARRAY_SIZE-1; m++){
pix_mat[k][m] = pix_mat[k][m+1];
}
}
for (i=0; i<ARRAY_SIZE-1; i++){
pix_mat[i][ARRAY_SIZE-1] = line_buf[i][x];
}
ap_int<32> y_val = conv_rgb2y_soft(pix.data);
pix_mat[ARRAY_SIZE-1][ARRAY_SIZE-1] = y_val;
for (i=0; i<ARRAY_SIZE-2; i++){ // 行の入れ替え
line_buf[i][x] = line_buf[i+1][x];
}
line_buf[ARRAY_SIZE-2][x] = y_val;
// 使用する配列を決定する
int ano;
switch (RorL){
case LEFT_WEIGHT :
ano = LEFT_WEIGHT;
break;
case RIGHT_WEIGHT :
ano = RIGHT_WEIGHT;
break;
case L_R_WEIGHT :
if (k == 0)
ano = LEFT_WEIGHT;
else
ano = RIGHT_WEIGHT;
break;
default :
ano = L_R_WEIGHT;
break;
}
// Gabor filter の演算
for (j=0, val=0; j<ARRAY_SIZE; j++){
for (i=0; i<ARRAY_SIZE; i++){
val += gabor_weight[ano][j][i] * pix_mat[j][i];
}
}
val = val/256; // 256倍してあるので、1/256して戻す
if (val<0)
//val = -val; // 絶対値
val = 0; // マイナスの値を0に丸める
else if (val>255)
val = 255;
// Gabor filter・データの書き込み
gabor.data = (val<<16)+(val<<8)+val;
// 最初のARRAY_SIZE-1行とその他の行の最初のARRAY_SIZE-1列は無効データなので0とする
if (x<(ARRAY_SIZE-1) || y<(ARRAY_SIZE-1))
gabor.data = 0;
if (x==0 && y==0) // 最初のデータでは、TUSERをアサートする
gabor.user = 1;
else
gabor.user = 0;
if (x == (HORIZONTAL_PIXEL_WIDTH-1)) // 行の最後で TLAST をアサートする
gabor.last = 1;
else
gabor.last = 0;
outs << gabor; // AXI4-Stream へ出力
}
}
}
return(0);
}
// RGBからYへの変換
// RGBのフォーマットは、{8'd0, R(8bits), G(8bits), B(8bits)}, 1pixel = 32bits
// 輝度信号Yのみに変換する。変換式は、Y = 0.299R + 0.587G + 0.114B
// "YUVフォーマット及び YUV<->RGB変換"を参考にした。http://vision.kuee.kyoto-u.ac.jp/~hiroaki/firewire/yuv.html
// 2013/09/27 : float を止めて、すべてint にした
int conv_rgb2y_soft(int rgb){
int r, g, b, y_f;
int y;
b = rgb & 0xff;
g = (rgb>>8) & 0xff;
r = (rgb>>16) & 0xff;
y_f = 77*r + 150*g + 29*b; //y_f = 0.299*r + 0.587*g + 0.114*b;の係数に256倍した
y = y_f >> 8; // 256で割る
return(y);
}
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n0.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n0.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:30:44.679307 1273 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n0.bmp result:
score[0] = 0.992188
score[1] = 0
score[2] = 0
score[3] = 0
score[4] = 0
score[5] = 0
score[6] = 0.0390625
score[7] = 0
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n1.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n1.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:30:54.338976 1282 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n1.bmp result:
score[0] = 0
score[1] = 0.992188
score[2] = 0.0859375
score[3] = 0
score[4] = 0
score[5] = 0
score[6] = 0
score[7] = 0.210938
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n2.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n2.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:30:59.285208 1287 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n2.bmp result:
score[0] = 0.164062
score[1] = 0
score[2] = 0.210938
score[3] = 0
score[4] = 0
score[5] = 0
score[6] = 0
score[7] = 0.25
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n3.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n3.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:31:15.731254 1300 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n3.bmp result:
score[0] = 0
score[1] = 0
score[2] = 0
score[3] = 0.992188
score[4] = 0
score[5] = 0.164062
score[6] = 0
score[7] = 0
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n4.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n4.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:31:20.559351 1305 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n4.bmp result:
score[0] = 0
score[1] = 0
score[2] = 0.0390625
score[3] = 0
score[4] = 0.789062
score[5] = 0
score[6] = 0
score[7] = 0
score[8] = 0
score[9] = 0.5
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n5.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n5.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:31:26.883575 1310 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n5.bmp result:
score[0] = 0
score[1] = 0
score[2] = 0
score[3] = 0.125
score[4] = 0
score[5] = 0.992188
score[6] = 0
score[7] = 0
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n6.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n6.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:31:31.589458 1315 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n6.bmp result:
score[0] = 0
score[1] = 0
score[2] = 0
score[3] = 0
score[4] = 0.414062
score[5] = 0.125
score[6] = 0.585938
score[7] = 0
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n7.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n7.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:31:39.147933 1324 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n7.bmp result:
score[0] = 0
score[1] = 0.414062
score[2] = 0
score[3] = 0
score[4] = 0.289062
score[5] = 0
score[6] = 0
score[7] = 0.625
score[8] = 0
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n8.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n8.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:31:54.020833 1337 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n8.bmp result:
score[0] = 0
score[1] = 0
score[2] = 0
score[3] = 0.460938
score[4] = 0.5
score[5] = 0
score[6] = 0
score[7] = 0
score[8] = 0.335938
score[9] = 0
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_custom_op.xmodel ~/tensorflow2_example/handwriting/n9.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_custom_op.xmodel
image_path: /home/petalinux/tensorflow2_example/handwriting/n9.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1211 04:32:02.205699 1342 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/handwriting/n9.bmp result:
score[0] = 0
score[1] = 0.210938
score[2] = 0
score[3] = 0.375
score[4] = 0
score[5] = 0
score[6] = 0
score[7] = 0
score[8] = 0
score[9] = 0
# Copyright 2019 Xilinx Inc.
#
# Licensed 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.
# train_eval_line.py
# 2022/12/05 : modified by marsee
# 2022/12/08 : modified by marsee
import os, sys
import numpy as np
import tensorflow as tf
import argparse
print(tf.__version__)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
from curve_dataset import load_mnist
from tensorflow.keras import utils as np_utils
parser = argparse.ArgumentParser(description='tensorflow 2 custom op example.')
parser.add_argument('--eval_only', type=bool, default=False, help='if only evaluate model')
parser.add_argument('--quant', type=bool, default=False, help='if evluate quantized model')
args = parser.parse_args()
class Mylayer(tf.keras.layers.Layer):
def __init__(self, name="custom_layer", **kwargs):
super().__init__(name=name, **kwargs)
self.prelu = tf.keras.layers.PReLU()
def call(self, inputs, training=None, mask=None):
x = self.prelu(inputs)
return x
custom_objects = {"Mylayer": Mylayer}
def build_model():
inputs = tf.keras.Input((10, 56, 1))
x = tf.keras.layers.Conv2D(2, (5, 5), activation='relu')(inputs)
x = tf.keras.layers.MaxPooling2D((2, 2), (2, 2))(x)
my_model = Mylayer()
x = my_model(x)
x = tf.keras.layers.Flatten()(x)
x = tf.keras.layers.Dense(100, activation='relu')(x)
x = tf.keras.layers.Dense(3, activation='softmax')(x)
model = tf.keras.Model(inputs=inputs, outputs=x)
return model
def main():
#################################
###### build dataset
#################################
(x_train, y_train), (x_test, y_test) = load_mnist(normalize=False, flatten=False)
trainzip = list(zip(y_train, x_train))
np.random.shuffle(trainzip)
y_train, x_train = zip(*trainzip)
x_train = np.asarray(x_train)
y_train = np.asarray(y_train)
testzip = list(zip(y_test, x_test))
np.random.shuffle(testzip)
y_test, x_test = zip(*testzip)
x_test = np.asarray(x_test)
y_test = np.asarray(y_test)
x_train = x_train.reshape(x_train.shape[0], 10, 56, 1)
x_test = x_test.reshape(x_test.shape[0], 10, 56, 1)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
y_train = y_train.astype('int32')
y_test = y_test.astype('int32')
y_train = np_utils.to_categorical(y_train, 3)
y_test = np_utils.to_categorical(y_test, 3)
#################################
##### evaluate model
#################################
if args.eval_only:
if args.quant:
from tensorflow_model_optimization.quantization.keras import vitis_quantize
# quantized_model = tf.keras.models.load_model("./quantized/quantized.h5",
# custom_objects=custom_objects)
with vitis_quantize.quantize_scope():
quantized_model = tf.keras.models.load_model("./quantized/quantized_curve_line.h5",
custom_objects=custom_objects)
quantized_model.compile(optimizer="adam", loss="binary_crossentropy", metrics=["accuracy"])
eval_results = quantized_model.evaluate(x_test, y_test)
print("\n***************** Summary *****************")
print("Quantized model accuracy: ", eval_results[1])
return
loaded_model = tf.keras.models.load_model("./keras_curve_cnn2_line.h5",
custom_objects=custom_objects)
eval_results = loaded_model.evaluate(x_test, y_test)
print("\n***************** Summary *****************")
print("Float model accuracy: ", eval_results[1])
return
#################################
##### build model
#################################
model = build_model()
model.summary()
#################################
##### compile train
#################################
model.compile(optimizer="adam", loss="binary_crossentropy", metrics=["accuracy"])
model.fit(x_train, y_train, epochs=7, shuffle=True)
eval_results = model.evaluate(x_test, y_test)
print("\n***************** Summary *****************")
print("Trained float model accuracy: ", eval_results[1])
save_path = "./keras_curve_cnn2_line.h5"
print("Trained float model is saved in ", save_path)
model.save(save_path)
if __name__ == '__main__':
main()
# Copyright 2019 Xilinx Inc.
#
# Licensed 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.
# quantize_curve_line.py
# 2022/12/05 : modified by marsee
import os, sys
import numpy as np
import tensorflow as tf
from curve_dataset import load_mnist
from train_eval_line import custom_objects
print(tf.__version__)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
def main():
#################################
###### build dataset
#################################
(x_train, y_train), (x_test, y_test) = load_mnist(normalize=False, flatten=False)
testzip = list(zip(y_test, x_test))
np.random.shuffle(testzip)
y_test, x_test = zip(*testzip)
x_test = np.asarray(x_test)
y_test = np.asarray(y_test)
x_test = x_test.reshape(x_test.shape[0], 10, 56, 1)
x_test = x_test.astype('float32')
x_test /= 255
#################################
##### quantize model
#################################
loaded_model = tf.keras.models.load_model("./keras_curve_cnn2_line.h5",
custom_objects=custom_objects)
loaded_model.summary()
from tensorflow_model_optimization.quantization.keras import vitis_quantize
# do quantization
calib_dataset = x_test
quant_model = vitis_quantize.VitisQuantizer(loaded_model, custom_objects=custom_objects).quantize_model(
calib_dataset=calib_dataset,
add_shape_info=True)
# quant_model.summary()
q_save_path = os.path.join('./quantized', 'quantized_curve_line.h5')
quant_model.save(q_save_path)
print("\n***************** Summary *****************")
print("Quantized model is saved in ", q_save_path)
if __name__ == '__main__':
main()
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/straight_RED_rect0_00.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/straight_RED_rect0_00.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:27:18.511101 1328 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/straight_RED_rect0_00.bmp result:
score[0] = 0.957431
score[1] = 0.0289119
score[2] = 0.013657
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/straight_RED_rect0_01.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/straight_RED_rect0_01.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:27:29.567533 1341 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/straight_RED_rect0_01.bmp result:
score[0] = 0.546549
score[1] = 0.331499
score[2] = 0.121952
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/straight_RED_rect0_02.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/straight_RED_rect0_02.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:27:34.319573 1346 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/straight_RED_rect0_02.bmp result:
score[0] = 0.554841
score[1] = 0.432111
score[2] = 0.0130486
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/straight_RED_rect0_03.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/straight_RED_rect0_03.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:27:38.321185 1347 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/straight_RED_rect0_03.bmp result:
score[0] = 0.261528
score[1] = 0.710907
score[2] = 0.0275649
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/straight_RED_rect0_04.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/straight_RED_rect0_04.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:27:56.568765 1364 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/straight_RED_rect0_04.bmp result:
score[0] = 0.74834
score[1] = 0.214403
score[2] = 0.0372576
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/left_turn_RED_rect0_00.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/left_turn_RED_rect0_00.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:44:29.216714 2167 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/left_turn_RED_rect0_00.bmp result:
score[0] = 0.145139
score[1] = 0.835218
score[2] = 0.0196425
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/left_turn_RED_rect0_01.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/left_turn_RED_rect0_01.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:44:43.451567 2181 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/left_turn_RED_rect0_01.bmp result:
score[0] = 0.542674
score[1] = 0.422635
score[2] = 0.034692
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/left_turn_RED_rect0_02.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/left_turn_RED_rect0_02.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:44:52.339522 2190 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/left_turn_RED_rect0_02.bmp result:
score[0] = 0.960332
score[1] = 0.0289995
score[2] = 0.0106683
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/left_turn_RED_rect0_03.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/left_turn_RED_rect0_03.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:44:56.608392 2195 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/left_turn_RED_rect0_03.bmp result:
score[0] = 0.999186
score[1] = 0.000261046
score[2] = 0.000552634
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/left_turn_RED_rect0_04.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/left_turn_RED_rect0_04.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:45:00.314903 2196 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/left_turn_RED_rect0_04.bmp result:
score[0] = 0.89917
score[1] = 0.00605856
score[2] = 0.0947718
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/right_turn_RED_rect0_00.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/right_turn_RED_rect0_00.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:48:50.522089 2382 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/right_turn_RED_rect0_00.bmp result:
score[0] = 0.00135824
score[1] = 0.09522
score[2] = 0.903422
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/right_turn_RED_rect0_01.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/right_turn_RED_rect0_01.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:48:56.709283 2391 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/right_turn_RED_rect0_01.bmp result:
score[0] = 5.81789e-05
score[1] = 0.222687
score[2] = 0.777255
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/right_turn_RED_rect0_02.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/right_turn_RED_rect0_02.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:49:01.729414 2396 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/right_turn_RED_rect0_02.bmp result:
score[0] = 0.00404301
score[1] = 0.00666579
score[2] = 0.989291
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/right_turn_RED_rect0_03.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/right_turn_RED_rect0_03.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:49:16.835510 2409 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/right_turn_RED_rect0_03.bmp result:
score[0] = 0.0105812
score[1] = 0.0369319
score[2] = 0.952487
xilinx-k26-starterkit-20221:~/tensorflow2_example/deployment/cpp$ ./tf2_custom_op_graph_runner ~/tensorflow2_example/tf2_curve_line.xmodel ~/tensorflow2_example/right_turn_RED_rect0_04.bmp
model_file: /home/petalinux/tensorflow2_example/tf2_curve_line.xmodel
image_path: /home/petalinux/tensorflow2_example/right_turn_RED_rect0_04.bmp
WARNING: Logging before InitGoogleLogging() is written to STDERR
W1210 16:49:21.233657 2410 tool_function.cpp:171] [UNILOG][WARNING] The operator named custom_layer, type: Mylayer, is not defined in XIR. XIR creates the definition of this operator automatically. You should specify the shape and the data_type of the output tensor of this operation by set_attr("shape", std::vector<int>) and set_attr("data_type", std::string)
image file /home/petalinux/tensorflow2_example/right_turn_RED_rect0_04.bmp result:
score[0] = 0.124399
score[1] = 0.159732
score[2] = 0.715869
