Add to video-demo the ability to process photos as well as images

This commit is contained in:
Abdelrahman Mahmoud 2016-04-05 11:25:50 -04:00
parent 0d8c63f959
commit 549e446f21
1 changed files with 75 additions and 51 deletions

View File

@ -3,13 +3,14 @@
#include <chrono>
#include <fstream>
#include <opencv2/highgui/highgui.hpp>
#include <highgui.h>
#include <opencv2/imgproc/imgproc.hpp>
#include <boost/filesystem.hpp>
#include <boost/timer/timer.hpp>
#include <boost/program_options.hpp>
#include "VideoDetector.h"
#include "PhotoDetector.h"
#include "AffdexException.h"
#include "AFaceListener.hpp"
@ -20,13 +21,17 @@
using namespace std;
using namespace affdex;
/// <summary>
/// Project demos how to use the Affdex Windows SDK VideoDetector
/// </summary>
int main(int argsc, char ** argsv)
{
//Defaults, overridden by the command line parameters
std::map<boost::filesystem::path, bool> VIDEO_EXTS = { {boost::filesystem::path(".avi"), 1},
{boost::filesystem::path(".mov"), 1},
{boost::filesystem::path(".flv"), 1},
{boost::filesystem::path(".webm"), 1},
{boost::filesystem::path(".wmv"), 1},
{boost::filesystem::path(".mp4"), 1} };
affdex::path DATA_FOLDER;
affdex::path LICENSE_PATH;
affdex::path videoPath;
@ -34,12 +39,12 @@ int main(int argsc, char ** argsv)
bool draw_display = true;
bool loop = false;
unsigned int nFaces = 1;
int faceDetectorMode = (int)FaceDetectorMode::SMALL_FACES;
int faceDetectorMode = (int)FaceDetectorMode::LARGE_FACES;
const int precision = 2;
std::cerr.precision(precision);
std::cout.precision(precision);
namespace po = boost::program_options; // abbreviate namespace
po::options_description description("Project for demoing the Windows SDK VideoDetector class (processing video files).");
description.add_options()
@ -76,7 +81,7 @@ int main(int argsc, char ** argsv)
std::cerr << "For help, use the -h option." << std::endl << std::endl;
return 1;
}
// Parse and check the data folder (with assets)
if (!boost::filesystem::exists(DATA_FOLDER))
{
@ -87,25 +92,37 @@ int main(int argsc, char ** argsv)
}
try
{
//Initialize the video file detector
VideoDetector videoDetector(process_framerate, nFaces, (affdex::FaceDetectorMode) faceDetectorMode);
std::shared_ptr<Detector> detector;
//Initialize out file
boost::filesystem::path csvPath(videoPath);
boost::filesystem::path fileExt = csvPath.extension();
csvPath.replace_extension(".csv");
std::ofstream csvFileStream(csvPath.c_str());
if (!csvFileStream.is_open())
{
std::cerr << "Unable to open csv file " << csvPath << std::endl;
return 1;
}
std::cout << "Max num of faces set to: " << videoDetector.getMaxNumberFaces() << std::endl;
if (VIDEO_EXTS[fileExt]) // IF it is a video file.
{
detector = std::make_shared<VideoDetector>(process_framerate, nFaces, (affdex::FaceDetectorMode) faceDetectorMode);
}
else //Otherwise it's a photo
{
detector = std::make_shared<PhotoDetector>(nFaces, (affdex::FaceDetectorMode) faceDetectorMode);
}
//VideoDetector videoDetector(process_framerate, nFaces, (affdex::FaceDetectorMode) faceDetectorMode);
std::cout << "Max num of faces set to: " << detector->getMaxNumberFaces() << std::endl;
std::string mode;
switch (videoDetector.getFaceDetectorMode())
switch (detector->getFaceDetectorMode())
{
case FaceDetectorMode::LARGE_FACES:
mode = "LARGE_FACES";
@ -116,67 +133,74 @@ int main(int argsc, char ** argsv)
default:
break;
}
std::cout << "Face detector mode set to: " << mode << std::endl;
shared_ptr<PlottingImageListener> listenPtr(new PlottingImageListener(csvFileStream, draw_display));
//Activate all the detectors
videoDetector.setDetectAllEmotions(true);
videoDetector.setDetectAllExpressions(true);
videoDetector.setDetectAllEmojis(true);
videoDetector.setDetectGender(true);
videoDetector.setDetectGlasses(true);
//Set the location of the data folder and license file
videoDetector.setClassifierPath(DATA_FOLDER);
videoDetector.setLicensePath(LICENSE_PATH);
//Add callback functions implementations
videoDetector.setImageListener(listenPtr.get());
videoDetector.start(); //Initialize the detectors .. call only once
detector->setDetectAllEmotions(true);
detector->setDetectAllExpressions(true);
detector->setDetectAllEmojis(true);
detector->setDetectGender(true);
detector->setDetectGlasses(true);
detector->setClassifierPath(DATA_FOLDER);
detector->setLicensePath(LICENSE_PATH);
detector->setImageListener(listenPtr.get());
detector->start(); //Initialize the detectors .. call only once
do
{
shared_ptr<StatusListener> videoListenPtr = std::make_shared<StatusListener>();
videoDetector.setProcessStatusListener(videoListenPtr.get());
videoDetector.process(videoPath); //Process a video
//For each frame processed
while (videoListenPtr->isRunning())
detector->setProcessStatusListener(videoListenPtr.get());
if (VIDEO_EXTS[fileExt])
{
((VideoDetector *)detector.get())->process(videoPath); //Process a video
}
else
{
//videoPath is of type std::wstring on windows, but std::string on other platforms.
cv::Mat img = cv::imread(std::string(videoPath.begin(), videoPath.end()));
// Create a frame
Frame frame(img.size().width, img.size().height, img.data, Frame::COLOR_FORMAT::BGR);
((PhotoDetector *)detector.get())->process(frame); //Process an image
}
while (!VIDEO_EXTS[fileExt] || videoListenPtr->isRunning())
{
if (listenPtr->getDataSize() > 0)
{
std::pair<Frame, std::map<FaceId, Face> > dataPoint = listenPtr->getData();
Frame frame = dataPoint.first;
std::map<FaceId, Face> faces = dataPoint.second;
//Draw on the GUI
if (draw_display)
{
listenPtr->draw(faces, frame);
}
std::cerr << "timestamp: " << frame.getTimestamp()
<< " cfps: " << listenPtr->getCaptureFrameRate()
<< " pfps: " << listenPtr->getProcessingFrameRate()
<< " faces: "<< faces.size() << endl;
//Output metrics to file
listenPtr->outputToFile(faces, frame.getTimestamp());
}
}
} while(loop);
videoDetector.stop();
detector->stop();
csvFileStream.close();
std::cout << "Output written to file: " << csvPath << std::endl;
}
catch (AffdexException ex)
{
std::cerr << ex.what();
}
return 0;
}