lua5.2's error: multiple Lua VMs detected - lua

I use 5.2 for learning recently, what I want to try like this:
Step 1, build a c module for lua:
#include "lua.h"
#include "lauxlib.h"
#include "lualib.h"
#include <stdlib.h>
static int add(lua_State *L) {
int x = luaL_checkint(L, -2);
int y = luaL_checkint(L, -1);
lua_pushinteger(L, x + y);
return 1;
}
static const struct luaL_Reg reg_lib[] = {
{"add", add}
};
int luaopen_tool(lua_State *L) {
luaL_newlib(L, reg_lib);
lua_setglobal(L, "tool");
return 0;
}
I compile and link it with liblua.a, and I'm sure it works well in lua script like "require("tool") tool.add(1, 2)"
Step 2, I write another C program that wants to require my c module in step 1 like this:
#include "lua.h"
#include "lauxlib.h"
#include "lualib.h"
#include <stdlib.h>
int main(int argc, char* const argv[]) {
lua_State *L = luaL_newstate();
luaL_requiref(L, "base", luaopen_base, 1);
luaL_requiref(L, "package", luaopen_package, 1);
lua_getglobal(L, "require");
if (!lua_isfunction(L, -1)) {
printf("require not found\n");
return 2;
}
lua_pushstring(L, "tool");
if (lua_pcall(L, 1, 1, 0) != LUA_OK) {
printf("require_fail=%s\n", lua_tostring(L, -1));
return 3;
}
lua_getfield(L, -1, "add");
lua_pushinteger(L, 2);
lua_pushinteger(L, 3);
lua_pcall(L, 2, 1, 0);
int n = luaL_checkint(L, -1);
printf("n=%d\n", n);
return 0;
}
I also compile & link with liblua.a, but error occurs when I run it:
"require_fail=multiple Lua VMs detected"
Someone's blog said that in lua5.2, you should link c module and c host program both dynamicly, but not staticly.
is there someone that has the same problem, or is there somehing wrong in my code, thanks.
NOTICE:
the problem has been solved by compile main program with -Wl,-E, thanks a lot for all your help ^^.

Don't link your C module with liblua.a when you create a .so from it. For examples, see my page of Lua libraries: http://www.tecgraf.puc-rio.br/~lhf/ftp/lua/ . You can link liblua.a statically into your main program but you have to export its symbols by adding -Wl,-E at link time. That's how the Lua interpreter is built in Linux.

Related

Having issues with opencv/dnn with caffe model

I am trying to make use of openpose example in opencv using caffe model and opencv/dnn.hpp
tutorial I have been following - https://www.learnopencv.com/deep-learning-based-human-pose-estimation-using-opencv-cpp-python/
we require 2 files for the network as said in the tutorial :
1 - prototxt - https://github.com/spmallick/learnopencv/blob/master/OpenPose/pose/coco/pose_deploy_linevec.prototxt
2 - caffe model - posefs1.perception.cs.cmu.edu/OpenPose/models/pose/coco/pose_iter_440000.caffemodel
ros node that I made following the tutorial :
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/dnn/dnn.hpp>
#include <sensor_msgs/image_encodings.h>
#include <ros/ros.h>
#include <image_transport/image_transport.h>
#include <cv_bridge/cv_bridge.h>
#include <iostream>
using namespace std;
using namespace cv;
using namespace cv::dnn;
static const std::string OPENCV_WINDOW = "Image window";
#define COCO
#ifdef COCO
const int POSE_PAIRS[17][2] =
{
{1,2}, {1,5}, {2,3},
{3,4}, {5,6}, {6,7},
{1,8}, {8,9}, {9,10},
{1,11}, {11,12}, {12,13},
{1,0},{0,14},
{14,16}, {0,15}, {15,17}
};
static const std::string protoFile = "pose/coco/pose_deploy_linevec.prototxt";
static const std::string weightsFile = "pose/coco/pose_iter_440000.caffemodel";
int nPoints = 18;
#endif
class ImageConverter
{
ros::NodeHandle nh_;
image_transport::ImageTransport it_;
image_transport::Subscriber image_sub_;
public:
ImageConverter()
: it_(nh_)
{
image_sub_ = it_.subscribe("/zed/rgb/image_raw_color", 1, &ImageConverter::imageCb, this);
}
~ImageConverter()
{
cv::destroyWindow(OPENCV_WINDOW);
}
void imageCb(const sensor_msgs::ImageConstPtr& msg)
{
cv_bridge::CvImagePtr cv_ptr;
try
{
cv_ptr = cv_bridge::toCvCopy(msg, sensor_msgs::image_encodings::BGR8);
}
catch (cv_bridge::Exception& e)
{
ROS_ERROR("cv_bridge exception: %s", e.what());
return;
}
if (cv_ptr->image.rows > 60 && cv_ptr->image.cols > 60)
detect_people(cv_ptr->image);
cv::waitKey(3);
}
void detect_people(cv::Mat msg)
{
int inWidth = msg.cols;
int inHeight = msg.rows;
float thresh = 0.1;
cv::Mat frame;
msg.copyTo(frame);
cv::Mat frameCopy = frame.clone();
int frameWidth = frame.cols;
int frameHeight = frame.rows;
cv::dnn::Net net = cv::dnn::readNetFromCaffe("pose_deploy_linevec.prototxt" ,"pose_iter_440000.caffemodel");
cv::Mat inpBlob = blobFromImage(frame, 1.0/255, cv::Size(inWidth, inHeight), cv::Scalar(0, 0, 0), false, false);
net.setInput(inpBlob);
cv::Mat output = net.forward();
int H = output.size[2];
int W = output.size[3];
std::vector<cv::Point> points(nPoints);
for (int n=0; n < nPoints; n++)
{
// Probability map of corresponding body's part.
cv::Mat probMap(H, W, CV_32F, output.ptr(0,n));
cv::Point2f p(-1,-1);
cv::Point maxLoc;
double prob;
cv::minMaxLoc(probMap, 0, &prob, 0, &maxLoc);
if (prob > thresh)
{
p = maxLoc;
p.x *= (float)frameWidth / W ;
p.y *= (float)frameHeight / H ;
cv::circle(frameCopy, cv::Point((int)p.x, (int)p.y), 8, Scalar(0,255,255), -1);
cv::putText(frameCopy, cv::format("%d", n), cv::Point((int)p.x, (int)p.y), cv::FONT_HERSHEY_COMPLEX, 1, cv::Scalar(0, 0, 255), 2);
}
points[n] = p;
}
int nPairs = sizeof(POSE_PAIRS)/sizeof(POSE_PAIRS[0]);
for (int n = 0; n < nPairs; n++)
{
// lookup 2 connected body/hand parts
Point2f partA = points[POSE_PAIRS[n][0]];
Point2f partB = points[POSE_PAIRS[n][1]];
if (partA.x<=0 || partA.y<=0 || partB.x<=0 || partB.y<=0)
continue;
cv::line(frame, partA, partB, cv::Scalar(0,255,255), 8);
cv::circle(frame, partA, 8, cv::Scalar(0,0,255), -1);
cv::circle(frame, partB, 8, cv::Scalar(0,0,255), -1);
}
cv::imshow("Output-Skeleton", frame);
}
};
int main(int argc, char** argv)
{
ros::init(argc, argv, "image_converter");
ros::NodeHandle nh_;
ros::Publisher pub;
ImageConverter ic;
ros::spin();
return 0;
}
The code is compiled without any errors, but while I run the code it gives the following error msg :
I get the following error when I run the node
error - OpenCV Error: Unspecified error (FAILED: fs.is_open(). Can't open "pose_deploy_linevec.prototxt") in ReadProtoFromTextFile, file /tmp/binarydeb/ros-kinetic-opencv3-3.3.1/modules/dnn/src/caffe/caffe_io.cpp, line 1119
terminate called after throwing an instance of 'cv::Exception'
what(): /tmp/binarydeb/ros-kinetic-opencv3-3.3.1/modules/dnn/src/caffe/caffe_io.cpp:1119: error: (-2) FAILED: fs.is_open(). Can't open "pose_deploy_linevec.prototxt" in function ReadProtoFromTextFile
Aborted (core dumped)
please help me solve this issue.
This issue is probably with the windows users only.
Solve the issue by:
Using/Adding absolute path when calling prototxt file.
Add the extension too. For Example:
"pose/coco/pose_deploy_linevec.prototxt.txt"
Spent 3 hours debugging this myself. Hope it helped someone else.
You are selecting the wrong file path.
Just replace this line:
static const std::string protoFile = "pose/coco/pose_deploy_linevec.prototxt";
with the path of the prototxt file in your laptop like this:
static const std::string protoFile = "C:/Users/lenovo/Desktop/learnopencv-master/OpenPose/pose/coco/pose_deploy_linevec.prototxt";

How to get list element from rosparam parameter from the cmd?

I loaded this yaml file:
num_boxes: 1
boxes: [[x: 0.349, y: 0.213, z: 0.117]]
with rosparam load my_config.yaml then I can do rosparam get boxes and get:
- - {x: 0.349}
- {y: 0.213}
- {z: 0.117}
But how can I access only the first list or elements in the second list? I tried boxes[0], boxes(0) and boxes{0} but nothing worked.
This is an old question, but I'm answering it because it's still useful to put an answer here.
Using the example given, boxes is a list of lists (YAML syntax). From the rosparam command line tool, we can't process rosparam get /boxes any further (excepting grep and regex). In common use, we can access the Parameter Server in C++ / Python.
# Python
>>> import rospy
>>> boxes = rospy.get_param("/boxes"); boxes
[[{'x': 0.349}, {'y': 0.213}, {'z': 0.117}]]
# Boxes is a python list/array of list of dicts
>>> boxes[0]
[{'x': 0.349}, {'y': 0.213}, {'z': 0.117}]
>>> boxes[0][0]
{'x': 0.349}
// C++
#include <vector>
#include <ros/ros.h>
// yaml_list: [1, 2]
ros::NodeHandle *nh;
std::vector<int> yaml_list;
double x = 0.0;
int main(int argc, char** argv){
ros::init(argc, argv, "get_params_node");
nh = new ros::NodeHandle("");
nh->getParam("/yaml_list", yaml_list);
x = yaml_list[0];
// ...
return 0;
}
To have deeper data structures, such as vectors of vectors of maps, you have to use the xmlrpcpp (.h/.cpp) interface.
// C++
#include <vector>
#include <map>
#include <string>
#include <ros/ros.h>
#include <xmlrpcpp/XmlRpcValue.h> // catkin component
ros::NodeHandle *nh;
XmlRpc::XmlRpcValue boxes; // std::vector<std::vector<std::map<std::string,double>>>
double x = 0.0;
int i = 0;
double point[3] = {0};
int main(int argc, char** argv){
ros::init(argc, argv, "get_params_node");
nh = new ros::NodeHandle("");
nh->getParam("/boxes", boxes);
if(boxes.getType() == XmlRpc::XmlRpcValue::Type::TypeArray && boxes.size() > 0){
// boxes[0] is a 'TypeArray' aka vector
if(boxes[0].getType() == XmlRpc::XmlRpcValue::Type::TypeArray && boxes[0].size() > 0){
// boxes[0][0] is a 'TypeStruct' aka map
if(boxes[0][0].getType() == XmlRpc::XmlRpcValue::Type::TypeStruct && boxes[0][0].hasMember("x")){
x = double(boxes[0][0]["x"]);
for(XmlRpc::XmlRpcValue::iterator it = boxes[0][0].begin(); it != boxes[0][0].end(); ++it){
point[i++] = double(*it);
}
}
}
}
// ...
return 0;
}
In standard use, the <rosparam> XML tag keeps the same YAML or rosparam command line syntax.

How can I store a big matrix within the .cc file?

I am currently working on a Computer Vision / Machine Learning project for university. Sadly, they only allow us to upload one single file and restrict the computation time too much. Hence I need to compute the matrices on my machine and store them in the same file as the code (22500 rows, 1 col and 100 rows + 22500 col and 100 rows + 1 col). I already found a way to export the data (link), but I'm not sure how to initialize the matrix.
What I've tried
#include <opencv/cv.h>
#include <opencv/highgui.h>
int main(int argc, char const *argv[])
{
float data[10] = {1,2,3,4,5,7,8,9,10,11};
cv::Mat A;
// Something is wrong with this line
A = cv::Mat(1, 10, cv::CV_32FC1, data);
return 0;
}
When I compile it, I get:
main.cc: In function ‘int main(int, const char**)’:
main.cc:10:16: error: expected primary-expression before ‘(’ token
A = cv::Mat(1, 10, cv::CV_32FC1, data);
^
In file included from /usr/include/opencv2/core/core_c.h:47:0,
from /usr/include/opencv/cv.h:63,
from main.cc:1:
main.cc:10:28: error: expected unqualified-id before ‘(’ token
A = cv::Mat(1, 10, cv::CV_32FC1, data);
^
Second try
#include <opencv/cv.h>
#include <opencv/highgui.h>
int main(int argc, char const *argv[])
{
float dataHeaderMat1[10] = {1,2,3,4,5,7,8,9,10,11};
cv::Mat matrix1;
// Something is wrong with this line
cv::cvInitMatHeader( &matrix1, 10, 1, CV_64FC1, dataHeaderMat1);
return 0;
}
gives
main.cc:10:5: error: ‘cvInitMatHeader’ is not a member of ‘cv’
cv::cvInitMatHeader( &matrix1, 10, 1, CV_64FC1, dataHeaderMat1);
^
The following works to declare and initialize a matrix:
#include <opencv/cv.h>
#include <opencv/highgui.h>
int main(int argc, char const *argv[])
{
float data[10] = {1,2,3,4,5,7,8,9,10,11};
cv::Mat A;
// Something is wrong with this line
A = cv::Mat(1, 10, CV_32FC1, data);
return 0;
}
However, I'm not too sure if this is the best way for big arrays.
You can try to save image to header file, like this:
#include <iostream>
#include <vector>
#include <string>
#include <fstream>
#include <opencv2/opencv.hpp>
using namespace cv;
using namespace std;
// uncomment for test
//#include "image.h"
int main(int argc, char **argv)
{
// This part creates header file from image.
Mat img=imread("D:\\ImagesForTest\\lena.jpg");
int w=img.cols;
int h=img.rows;
int channels=img.channels();
ofstream os("image.h");
os << "int rows=" << h << ";" << endl;
os << "int cols=" << w << ";" << endl;
os << "unsigned char d[]={" << endl;
for(int i=0;i<h;++i)
{
for(int j=0;j<w;++j)
{
if(i!=(w-1) || j!=(h-1))
{
Vec3b b=img.at<Vec3b>(i,j);
os << format("0x%02x,",b[0]);
os << format("0x%02x,",b[1]);
os << format("0x%02x,",b[2]);
}
}
}
Vec3b b=img.at<Vec3b>(w-1,h-1);
os << format("0x%02x,",b[0]);
os << format("0x%02x,",b[1]);
os << format("0x%02x",b[2]);
os << endl << "};" << endl;
os << "Mat I=Mat(rows,cols,CV_8UC3,d);" << endl;
os.close();
// To test uncomment commented part of code and comment uncommented.
// uncomment for test
/*
namedWindow("I");
imshow("I",I);
waitKey();
return 0;
*/
}
But be careful, not all IDEs likes such large files.

Reading analog inputs fast in beaglebone black

I need to read all 7 analog pins in the BBB every 5 milliseconds. I'm doing so with the following C code:
void main(){
char value_str[7];
long int value_int = 0;
FILE* f0 = fopen("/sys/bus/iio/devices/iio:device0/in_voltage0_raw", "r");
while(1){
fread(&value_str, 6, 6, f0);
value_int = strtol(value_str,NULL,0);
printf("0 %li\n", value_int);
fflush(stdout);
usleep(5000);
rewind(f0);
}
Hoever, the cpu usage goes up really high (20%). Is there any way to read the analog inputs differently so that it doesn't use as much CPU? Someone suggested "DMA" but I'm completely lost on that regard...
Any help will be appreciated.
This thread in the BBB forum was very useful and I ended up using libpruio. It uses the PRU to read the Beagle's built in io pins and analog to digital converters. The code I ended up using:
#include <stdio.h>
#include <unistd.h>
#include "pruio_c_wrapper.h"
#include "pruio_pins.h"
int main(int argc, const char *argv[]) {
PruIo *io = pruio_new(0, 0x98, 0, 1);
if (io->Errr) {
printf("Initialisation failed (%s)\n", io->Errr);
return 1;
}
if(pruio_config(io, 0, 0x1FE, 0, 4, 0)){
printf("Config failed (%s)\n", io->Errr);
return 1;
}
while(1){
printf"\r%12o %12o %12o %12o %4X %4X %4X %4X %4X %4X %4X %4X\n"
, io->Gpio[0].Stat, io->Gpio[1].Stat, io->Gpio[2].Stat, io->Gpio[3].Stat
, io->Value[1], io->Value[2], io->Value[3], io->Value[4], io->Value[5]
, io->Value[6], io->Value[7], io->Value[8]);
usleep(1000);
}
pruio_destroy(io);
return 0;
}
I suggest you use the PRU. It's very fast!
This should get you started->
http://www.element14.com/community/community/knode/single-board_computers/next-gen_beaglebone/blog/2013/08/04/bbb--high-speed-data-acquisition-and-web-based-ui

Can OpenCV be developed using C++ and C together

Can OpenCV be developed using C++ and C together? Following is the program where i meet this problem.It is coded with C and works well.But if I use
cv::Mat saveImage=cv::imread("D:\\opencvStudy\\opencv_test\\TaskDemo\\TaskDemo\\save.jpg");
cv::imshow("save",saveImage);
to replace
IplImage* saveImge =cvLoadImage("D:\\opencvStudy\\opencv_test\\TaskDemo\\TaskDemo\\save.jpg");
cvShowImage("saveimage",saveImge);
I met this:
Unhandled exception at 0x75709673 in TaskDemo.exe: Microsoft C++ exception: cv::Exception at memory location 0x0039ea0c..
Following is the whole program. Hope anyone could help me .Thanks very much
#include<opencv/cv.h>
#include<opencv/highgui.h>
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/video/tracking.hpp"
#include <iostream>
#include <ctype.h>
CvPoint pt1 = cvPoint(0,0);
CvPoint pt2 = cvPoint(0,0);
bool is_selecting = false;
void cvMouseCallback(int mouseEvent,int x,int y,int flags,void* param)
{
switch(mouseEvent)
{
case CV_EVENT_LBUTTONDOWN:
pt1 = cvPoint(x,y);
pt2 = cvPoint(x,y);
is_selecting = true;
break;
case CV_EVENT_MOUSEMOVE:
if(is_selecting)
pt2 = cvPoint(x,y);
break;
case CV_EVENT_LBUTTONUP:
pt2 = cvPoint(x,y);
is_selecting = false;
break;
}
return;
}
int main(int argc,char* argv[])
{
char img_path[80] = "D:\\opencvStudy\\pic\\boldt.jpg";
char save_path[80] = "save.jpg";
char* window = "img";
IplImage* img = cvLoadImage(img_path);
IplImage* img_show = cvCloneImage(img);
cvNamedWindow(window,CV_WINDOW_AUTOSIZE);
cvSetMouseCallback(window,cvMouseCallback);
char text[80];
CvFont font;
cvInitFont(&font,CV_FONT_HERSHEY_PLAIN,1.0,1.0);
while(true)
{
cvCopy(img,img_show);
cvRectangle(img_show,pt1,pt2,cvScalar(255,255,255));
sprintf(text,"roi = cvRect(%d,%d,%d,%d)",pt1.x,pt1.y,std::abs(pt2.x-pt1.x),std::abs(pt2.y-pt1.y));
cvPutText(img_show,text,cvPoint(10,20),&font,cvScalar(0,0,255));
cvShowImage(window,img_show);
char key = cvWaitKey(10);
if (key='r')
{
cvSetImageROI(img,cvRect(pt1.x,pt1.y,std::abs(pt2.x-pt1.x),std::abs(pt2.y-pt1.y)));
cvSaveImage(save_path,img);
cvResetImageROI(img);
IplImage* saveImge = cvLoadImage("D:\\opencvStudy\\opencv_test\\TaskDemo\\TaskDemo\\save.jpg");
cvShowImage("saveimage",saveImge);
//cv::Mat saveImage=cv::imread("D:\\opencvStudy\\opencv_test\\TaskDemo\\TaskDemo\\save.jpg");
//cv::imshow("save",saveImage);
}
else if(key==27)
break;
}
cvReleaseImage(&img);
cvReleaseImage(&img_show);
return 0;
}
Update
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/core/core.hpp"
#include <iostream>
#include <ctype.h>
using namespace std;
using namespace cv;
cv::Point pt1=Point(0,0);
cv::Point pt2=Point(0,0);
bool is_selecting=false;
static void onMouse(int mouseEvent,int x,int y,int flags,void* param)
{
switch(mouseEvent)
{
case CV_EVENT_LBUTTONDOWN://CV_EVENT_LBUTTONDOWN
pt1 = Point(x,y);
pt2 = Point(x,y);
is_selecting = true;
break;
case CV_EVENT_MOUSEMOVE:
if(is_selecting)
pt2 = Point(x,y);
break;
case CV_EVENT_LBUTTONUP:
pt2 = Point(x,y);
is_selecting = false;
break;
}
}
int main()
{
cv::Mat img=cv::imread("D:\\opencvStudy\\pic\\boldt.jpg");
cv::Mat img_show=img.clone();
cv::namedWindow("Task2",0);
cv::setMouseCallback("Task2",onMouse,0);
char text[80];
while(true)
{
img.copyTo(img_show);
cv::rectangle(img_show,pt1,pt2,cv::Scalar(255,255,255));
sprintf(text,"roi = cvRect(%d,%d,%d,%d)",pt1.x,pt1.y,std::abs(pt2.x-pt1.x),std::abs(pt2.y-pt1.y));
cv::putText(img_show,text,cvPoint(10,20),10,10,Scalar(0,0,255));
cv::imshow("Task2",img_show);
char key=cv::waitKey(10);
if (key='r')
{
cv::Mat save=img(cv::Rect(pt1.x,pt1.y,std::abs(pt2.x-pt1.x),std::abs(pt2.y-pt1.y)));
cv::imwrite("save.jpg",save);
cv::Mat saveImage=cv::imread("D:\\opencvStudy\\opencv_test\\TaskDemo\\TaskDemo\\save.jpg");
cv::imshow("save",saveImage);
}
else if(key==27)
break;
}
return 0;
}
The problem is with conflicting headers. You should not include old and new at the same time like this:
#include<opencv/cv.h>
#include<opencv/highgui.h>
#include "opencv2/highgui/highgui.hpp"
Instead, include the new headers only, like this:
#include <opencv2/core/core.hpp>
#include <opencv2/highgui/highgui.hpp>
The reason for this is that the C++ headers also include the C interface.
UPDATE:
Of course if you use cv::Mat then you need to use the new functions for that as well.
CvCopy becomes cv::copy, there is no cvReleaseImage needed, cvRectangle becomes cv::rectangle, cvSetImageROI is img(rect), cvPoint becomes cv::Point cvRect becomes cv::Rect
cvCloneImage becomes img.clone() and so on...
Or you need some conversion from cv::Mat to IplImage to use the old functions. There are some questions on SO dealing with converting back-and-forth.
UPDATE2:
It is best to remove all cv:: since you are using using namespace cv.
Then check for functions and variables left with the old cv prefix, and remove those prefixes. (I've still found one: cvPoint).
Hang on, you are using char text[80]!
I think that is again a source of problems. It is best to switch it to
#include <sstream>
//...
stringstream ss;
ss << "Rect(" << pt1.x << ", " << ... << ")";
//...
putText(img_show, ss.str(), Point(10,20), FONT_HERSHEY_PLAIN, 1, Scalar(0, 0, 255) );
Note that the font (I use) is not 10, but FONT_HERSHEY_PLAIN (or some other FONT_ constant).

Resources