mmr/tkDNN
Archived
1
0
Fork 0
Code Issues 46 Pull Requests 12 Releases 4 Activity

Merge branch 'class' of https://github.com/ceccocats/tkDNN into class

Browse Source
This commit is contained in:
Francesco Gatti
2019-03-06 16:35:02 +01:00
parent d631169821 61b6621d2c
commit 1c4aa3c5d7
4 changed files with 321 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
demo/demo/demo.cpp
+59 -2
View File
@@ -10,6 +10,11 @@
#include <opencv2/imgproc/imgproc.hpp>
#include "Yolo3Detection.h"
#include "send.h"
#define MAX_DETECT_SIZE 100
bool gRun;
@@ -30,9 +35,20 @@ int main(int argc, char *argv[]) {
char *input = "../demo/yolo_test.mp4";
if(argc > 2)
input = argv[2];
char *pmatrix = "/home/classfog1/repos/MASA_server/pmatrix/proj_matrix_20937.txt";
if(argc > 3)
pmatrix = argv[3];
/*CAMID*/
int CAM_IDX = 0;
if(argc > 4)
CAM_IDX = atoi(argv[4]);
bool to_show = true;
if(argc > 5)
to_show = atoi(argv[5]);
tk::dnn::Yolo3Detection yolo;
yolo.init(net);
yolo.thresh = 0.25;
gRun = true;
@@ -44,9 +60,21 @@ int main(int argc, char *argv[]) {
cv::Mat frame;
cv::Mat dnn_input;
cv::namedWindow("detection", cv::WINDOW_NORMAL);
cv::namedWindow("detection", cv::WINDOW_NORMAL);
/*projection matrix*/
float* proj_matrix = (float*) malloc(9*sizeof(float));
int proj_matrix_read = 0;
/*socket*/
int sock;
int socket_opened = 0;
struct obj_coords *coords = (struct obj_coords*)malloc(MAX_DETECT_SIZE*sizeof(struct obj_coords));
while(gRun) {
cap >> frame;
if(!frame.data) {
continue;
@@ -57,24 +85,53 @@ int main(int argc, char *argv[]) {
// TODO: async infer
yolo.update(dnn_input);
int coord_i = 0;
int num_detected = yolo.detected.size();
if (num_detected > MAX_DETECT_SIZE)
num_detected = MAX_DETECT_SIZE;
if(proj_matrix_read == 0)
read_projection_matrix(proj_matrix, proj_matrix_read, pmatrix);
/*printf("%f %f %f \n%f %f %f\n %f %f %f\n\n", proj_matrix[0],proj_matrix[1],
proj_matrix[2],proj_matrix[3],proj_matrix[4],proj_matrix[5],
proj_matrix[6],proj_matrix[7],proj_matrix[8]);*/
// draw dets
for(int i=0; i<yolo.detected.size(); i++) {
for(int i=0; i<num_detected; i++) {
tk::dnn::box b = yolo.detected[i];
int x0 = b.x;
int x1 = b.x + b.w;
int y0 = b.y;
int y1 = b.y + b.h;
int obj_class = b.cl;
if(obj_class == 0 /*person*/ || obj_class == 1/*bicycle*/ || obj_class == 2/*car*/
|| obj_class == 3/*motorbike*/ || obj_class == 5/*bus*/)
{
convert_coords(coords, coord_i,x0+b.w/2, y1,obj_class, proj_matrix);
coord_i++;
}
float prob = b.prob;
std::cout<<obj_class<<" ("<<prob<<"): "<<x0<<" "<<y0<<" "<<x1<<" "<<y1<<"\n";
cv::rectangle(frame, cv::Point(x0, y0), cv::Point(x1, y1), yolo.colors[obj_class], 2);
}
send_client_dummy(coords, coord_i, sock, socket_opened, CAM_IDX);
if (to_show)
{
cv::imshow("detection", frame);
cv::waitKey(1);
}
}
free(coords);
free(proj_matrix);
std::cout<<"detection end\n";
return 0;
}
include/Yolo3Detection.h
+3 -1
View File
@@ -30,6 +30,8 @@ class Yolo3Detection {
cv::Mat imageF;
cv::Mat bgr[3];
public:
int classes = 0;
int num = 0;
@@ -49,7 +51,7 @@ class Yolo3Detection {
* @return Success of the initialization
*/
bool init(std::string tensor_path);
void addBorders(cv::Mat &imageORIG, cv::Mat &imageWBorders, int &top, int &left);
void update(cv::Mat &frame);
};
include/send.h
+189
View File
@@ -0,0 +1,189 @@
#ifndef SEND_H
#define SEND_H
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/time.h>
#include <sys/socket.h> //socket
#include <arpa/inet.h> //inet_addr
#include <unistd.h> //write
struct obj_coords
{
float LAT;
float LONG;
float cl;
};
char *serialize_coords(struct obj_coords *c, int obj_n, int CAM_IDX)
{
char *buffer = (char *)malloc(100000 * sizeof(char));
struct timeval tv;
gettimeofday(&tv, NULL);
unsigned long long t_stamp_ms = (unsigned long long)(tv.tv_sec) * 1000 + (unsigned long long)(tv.tv_usec) / 1000;
sprintf(buffer, "m %lld %d %d ", t_stamp_ms, CAM_IDX, obj_n);
int i;
for (i = 0; i < obj_n; i++)
sprintf(buffer + strlen(buffer), "%.9f %.9f %.0f ", c[i].LAT, c[i].LONG, c[i].cl);
//printf("%s\n", buffer);
return buffer;
}
struct obj_coords *deserialize_coords(char *buffer, int *obj_n)
{
int cam_id;
unsigned long long t_stamp_ms;
char type_of_m;
int consumed_chars = 0;
char shifted_chars[100000];
sscanf(buffer, "%c %lld %d %d ", &type_of_m, &t_stamp_ms, &cam_id, obj_n);
sprintf(shifted_chars, "%c %lld %d %d ", type_of_m, t_stamp_ms, cam_id, *obj_n);
consumed_chars += strlen(shifted_chars);
//printf("%c %lld %d %d\n", type_of_m, t_stamp_ms, cam_id, *obj_n);
struct obj_coords *c = (struct obj_coords *)malloc(*obj_n * sizeof(struct obj_coords));
int i;
for (i = 0; i < *obj_n; i++)
{
sscanf(buffer + consumed_chars, "%f %f %f ", &c[i].LAT, &c[i].LONG, &c[i].cl);
sprintf(shifted_chars, "%.9f %.9f %.0f ", c[i].LAT, c[i].LONG, c[i].cl);
consumed_chars += strlen(shifted_chars);
//printf("%Lf %f %f \n", c[i].LAT, c[i].LONG, c[i].cl);
}
return c;
}
int map_class_coco_to_voc(int coco_class)
{
switch (coco_class)
{
case 0:
return 14; //person
case 1:
return 1; //bicycle
case 2:
return 6; //car
case 3:
return 13; //motorkbike
case 5:
return 5; //bus
}
return -1;
}
void convert_coords(struct obj_coords *coords, int i, int x, int y, int detected_class,float * proj_matrix)
{
float obj_x = x, obj_y = y, obj_z = 1;
float tmp_z = 0;
coords[i].LAT = proj_matrix[0] * obj_x + proj_matrix[1] * obj_y + proj_matrix[2] * obj_z;
coords[i].LONG = proj_matrix[3] * obj_x + proj_matrix[4] * obj_y + proj_matrix[5] * obj_z;
tmp_z = proj_matrix[6] * obj_x + proj_matrix[7] * obj_y + proj_matrix[8] * obj_z;
if(tmp_z != 0.0)
{
coords[i].LAT = coords[i].LAT / tmp_z;
coords[i].LONG = coords[i].LONG / tmp_z;
//printf("lat: %f, long %f\n", coords[i].LAT, coords[i].LONG);
}
else
printf("Division by 0 (tmp_z)\n");
coords[i].cl = map_class_coco_to_voc(detected_class);
}
void read_projection_matrix(float * proj_matrix, int &proj_matrix_read, char* path)
{
FILE *fp;
char *line = NULL;
size_t len = 0;
ssize_t read;
fp = fopen(path, "r");
if (fp == NULL)
exit(EXIT_FAILURE);
int i = 0;
while ((read = getline(&line, &len, fp)) != -1)
{
if (3 == sscanf(line, "%f %f %f", &proj_matrix[i*3+0], &proj_matrix[i*3+1], &proj_matrix[i*3+2]))
{
i++;
proj_matrix_read = 1;
}
}
free(line);
fclose(fp);
}
int open_socket(char *ip, int &sock, int &socket_opened)
{
struct sockaddr_in server;
sock = socket(AF_INET, SOCK_STREAM, 0);
if (sock == -1)
{
printf("Could not create socket");
}
puts("Socket created");
server.sin_addr.s_addr = inet_addr(ip);
server.sin_family = AF_INET;
server.sin_port = htons(8888);
/*connect to remote server*/
if (connect(sock, (struct sockaddr *)&server, sizeof(server)) < 0)
{
perror("connect failed. Error");
socket_opened = 0;
return 0;
}
puts("Connected\n");
socket_opened = 1;
return 1;
}
int send_client_dummy(struct obj_coords *coords, int n_coords, int &sock, int &socket_opened, int CAM_IDX)
{
/*serialize coords*/
char *message = serialize_coords(coords, n_coords, CAM_IDX);
/*open socket if not already opened*/
if (socket_opened == 0)
{
int res = open_socket("127.0.0.1",sock, socket_opened);
if(res)
printf("Socket opened!\n");
else
{
printf("Problem: socket NOT opened!\n");
return 0;
}
}
/*send message to server*/
if (send(sock, message, strlen(message), 0) < 0)
{
puts("Send failed");
socket_opened = 0;
}
free(message);
//close(sock);
return 1;
}
#endif /*SEND_H*/
src/Yolo3Detection.cpp
+70 -7
View File
@@ -57,6 +57,60 @@ bool Yolo3Detection::init(std::string tensor_path) {
return true;
}
void Yolo3Detection::addBorders(cv::Mat &imageORIG, cv::Mat &imageWBorders, int &top, int &left)
{
float net_ratio = float(netRT->input_dim.w)/float(netRT->input_dim.h);
float img_ratio = float(imageORIG.cols)/float(imageORIG.rows);
int bottom=0, right=0, diff= 0;
top=0, left=0;
//printf("%f %f\n", net_ratio, img_ratio);
if(net_ratio != img_ratio)
{
if(netRT->input_dim.w> netRT->input_dim.h)
{
if(img_ratio > net_ratio)
{
diff = std::abs((imageORIG.cols - net_ratio*imageORIG.rows)/net_ratio);
top = diff/2;
bottom = diff/2 + diff%2;
}
else
{
diff = std::abs(net_ratio*float(imageORIG.rows) - float(imageORIG.cols));
left = diff/2;
right = diff/2 + diff%2;
}
}
else
{
if(img_ratio < net_ratio)
{
diff = std::abs((imageORIG.cols - net_ratio*imageORIG.rows)/net_ratio);
left = diff/2;
right = diff/2 + diff%2;
}
else
{
diff = std::abs(net_ratio*float(imageORIG.rows) - float(imageORIG.cols));
top = diff/2;
bottom = diff/2 + diff%2;
}
}
}
//printf("%d %d %d %d %d \n", diff, top, bottom, left, right);
imageWBorders = imageORIG;
copyMakeBorder( imageORIG, imageWBorders, top, bottom, left, right, cv::BORDER_CONSTANT, (0,0,0) );
//printf("%d %d\n", imageWBorders.cols, imageWBorders.rows);
//const char* window_name = "borders";
//cv::namedWindow( window_name, cv::WINDOW_AUTOSIZE );
//imshow( window_name, imageWBorders );
//cv::waitKey(0);
}
void Yolo3Detection::update(cv::Mat &imageORIG) {
@@ -64,12 +118,21 @@ void Yolo3Detection::update(cv::Mat &imageORIG) {
std::cout<<"YOLO: NO IMAGE DATA\n";
return;
}
float xRatio = float(imageORIG.cols) / float(netRT->input_dim.w);
float yRatio = float(imageORIG.rows) / float(netRT->input_dim.h);
int top, left;
cv::Mat imageWBorders;
addBorders(imageORIG, imageWBorders, top, left);
resize(imageORIG, imageORIG, cv::Size(netRT->input_dim.w, netRT->input_dim.h));
float xRatio = float(imageWBorders.cols) / float(netRT->input_dim.w);
float yRatio = float(imageWBorders.rows) / float(netRT->input_dim.h);
resize(imageWBorders, imageORIG, cv::Size(netRT->input_dim.w, netRT->input_dim.h));
imageORIG.convertTo(imageF, CV_32FC3, 1/255.0);
//const char* window_name = "resize";
//cv::namedWindow( window_name, cv::WINDOW_AUTOSIZE );
///imshow( window_name, imageORIG );
//split channels
cv::split(imageF,bgr);//split source
@@ -127,10 +190,10 @@ void Yolo3Detection::update(cv::Mat &imageORIG) {
//cv::rectangle(image, cv::Point(x0, y0), cv::Point(x1, y1), colors[obj_class], 2);
// convert to image coords
x0 = xRatio*x0;
x1 = xRatio*x1;
y0 = yRatio*y0;
y1 = yRatio*y1;
x0 = xRatio*x0 - left;
x1 = xRatio*x1 - left;
y0 = yRatio*y0 - top;
y1 = yRatio*y1 - top;
tk::dnn::box res;
res.cl = obj_class;