tracking-by-detection icon indicating copy to clipboard operation
tracking-by-detection copied to clipboard

Running the demo

Open ashnarayan13 opened this issue 7 years ago • 8 comments

Hi,

I could set the path with the pkg-config for the dependencies and also did a make compile in the cpp/ folder.

I am not sure where the executable is built and how to run a demo with the project (is there a demo?). Or is this just a library and I have to link it to my project?

Thanks!

ashnarayan13 avatar Dec 20 '17 12:12 ashnarayan13

firstly, make comile,then make detect or make track or make detectAndTrack, finally, it output xxx.out file.

zg9uagfv avatar Dec 25 '17 08:12 zg9uagfv

I error run the demo. malu@malu1:~/tracking-by-detection/cpp/src/examples$ g++ -std=c++11 TrackDemo.cpp /tmp/ccxvm4i2.o:在函数‘track(boost::filesystem::path const&, std::string const&, std::string const&, int, bool)’中: TrackDemo.cpp:(.text+0x2a7):对‘boost::filesystem::path::parent_path() const’未定义的引用 TrackDemo.cpp:(.text+0x39f):对‘boost::filesystem::path::filename() const’未定义的引用 TrackDemo.cpp:(.text+0x615):对‘DetectionFileParser::parseOkutamaFile(std::basic_ifstream<char, std::char_traits >&)’未定义的引用 TrackDemo.cpp:(.text+0x63d):对‘DetectionFileParser::parseMOTFile(std::basic_ifstream<char, std::char_traits >&)’未定义的引用 TrackDemo.cpp:(.text+0x7ce):对‘PAOT::track(std::vector<Detection, std::allocator<Detection> > const&)’未定义的引用 /tmp/ccxvm4i2.o:在函数‘__static_initialization_and_destruction_0(int, int)’中: TrackDemo.cpp:(.text+0x14a0):对‘boost::system::generic_category()’未定义的引用 TrackDemo.cpp:(.text+0x14ac):对‘boost::system::generic_category()’未定义的引用 TrackDemo.cpp:(.text+0x14b8):对‘boost::system::system_category()’未定义的引用 TrackDemo.cpp:(.text+0x14ef):对‘boost::filesystem::path::parent_path() const’未定义的引用 /tmp/ccxvm4i2.o:在函数‘boost::filesystem::operator/(boost::filesystem::path const&, boost::filesystem::path const&)’中: TrackDemo.cpp:(.text.ZN5boost10filesystemdvERKNS0_4pathES3[ZN5boost10filesystemdvERKNS0_4pathES3]+0x37):对‘boost::filesystem::path::operator/=(boost::filesystem::path const&)’未定义的引用 /tmp/ccxvm4i2.o:在函数‘boost::filesystem::exists(boost::filesystem::path const&)’中: TrackDemo.cpp:(.text._ZN5boost10filesystem6existsERKNS0_4pathE[_ZN5boost10filesystem6existsERKNS0_4pathE]+0x20):对‘boost::filesystem::detail::status(boost::filesystem::path const&, boost::system::error_code*)’未定义的引用 /tmp/ccxvm4i2.o:在函数‘boost::filesystem::is_directory(boost::filesystem::path const&)’中: TrackDemo.cpp:(.text._ZN5boost10filesystem12is_directoryERKNS0_4pathE[_ZN5boost10filesystem12is_directoryERKNS0_4pathE]+0x20):对‘boost::filesystem::detail::status(boost::filesystem::path const&, boost::system::error_code*)’未定义的引用 /tmp/ccxvm4i2.o:在函数‘boost::filesystem::create_directories(boost::filesystem::path const&)’中: TrackDemo.cpp:(.text._ZN5boost10filesystem18create_directoriesERKNS0_4pathE[_ZN5boost10filesystem18create_directoriesERKNS0_4pathE]+0x19):对‘boost::filesystem::detail::create_directories(boost::filesystem::path const&, boost::system::error_code*)’未定义的引用 /tmp/ccxvm4i2.o:在函数‘boost::filesystem::current_path()’中: TrackDemo.cpp:(.text._ZN5boost10filesystem12current_pathEv[_ZN5boost10filesystem12current_pathEv]+0x19):对‘boost::filesystem::detail::current_path(boost::system::error_code*)’未定义的引用 /tmp/ccxvm4i2.o:在函数‘PAOT::PAOT()’中: TrackDemo.cpp:(.text._ZN4PAOTC2Ev[_ZN4PAOTC5Ev]+0x20):对‘vtable for PAOT’未定义的引用 /tmp/ccxvm4i2.o:在函数‘PAOT::~PAOT()’中: TrackDemo.cpp:(.text._ZN4PAOTD2Ev[_ZN4PAOTD5Ev]+0x13):对‘vtable for PAOT’未定义的引用 collect2: error: ld returned 1 exit status can you help me?

ml930310 avatar Dec 28 '17 08:12 ml930310

@ml930310 It is easier run demo by make compile and make track. It will build all the necessary links between the libraries and the target file for you.

jazzseow avatar Jan 03 '18 08:01 jazzseow

@jazzseow Thank you very much.You know how to set the downloaded data? Ssd model should be how to set? Thank you again.

ml930310 avatar Jan 03 '18 14:01 ml930310

@ml930310 Put seqmap file, which stores the relative path (from data/) to your detection data, under data/. The content of the seqmap is something like this:

../../dataset/MOT16/train/MOT16-02 ../../dataset/MOT16/train/MOT16-04 ../../dataset/MOT16/train/MOT16-05 ../../dataset/MOT16/train/MOT16-09 ../../dataset/MOT16/train/MOT16-10 ../../dataset/MOT16/train/MOT16-11 ../../dataset/MOT16/train/MOT16-13

Then run ./trackApp.out -s <seqmap file name> -m <model type> -f <detection format: either okutama or mot> , example ./trackApp.out -s seqmap.txt -m det -f mot. It will locate these folders and perform tracking using the det.txt with the path mentioned, e.g. ../../dataset/MOT16/train/MOT16-13/det/det.txt. I have only tried the tracking part, so I do not know much about setting the detection model. I guess you can try to read the demo codes under examples/ for more information.

But my way of doing this might be wrong. The values and results i got from the tracking are quite weird https://github.com/samuelmurray/tracking-by-detection/issues/5. Do you get this kind of result too?

jazzseow avatar Jan 04 '18 01:01 jazzseow

Hello, thank you for your reply, recently been preparing for the exam, did not promptly reply to you, I would like to ask you a question, I "make track" has been unsuccessful, has been unable to find a solution. The question is as follows:

1, malu @ malu1: ~ / tracking-by-detection / cpp $ make track (compile) CXX src / detector / SSDDetector.cpp In file included from src / detector / SSDDetector.h: 11: 0, from src / detector / SSDDetector.cpp: 1: /home/malu/caffe/include/caffe/caffe.hpp:7:26: fatal error: caffe / blob.hpp: No such file or directory #include "caffe / blob.hpp" ^ compilation terminated. make: *** [build / detector / SSDDetector.o] error 1

2, caffe.pc reads as follows:

Package Information for pkg-config

Path to where Caffe is installed

prefix = / home / malu / caffe

Path to where libraries are

libdir = $ {prefix} / build / lib

Path to where include files are

includedir = $ {prefix} / include / caffe

Name: Caffe Description: Deep learning framework by the BVLC Version: 1.0 Libs: -L $ {libdir} -lcaffe -lglog Cflags: -isystem $ {includedir} -isystem $ {prefix} / build / src #Cflags: -isystem $ {includedir} -isystem $ {prefix} / build / src

3, dlib.pc reads as follows: prefix = / usr / local exec_prefix = $ {prefix} libdir = $ {exec_prefix} / lib includedir = $ {prefix} / include

Name: dlib Description: Numerical and networking C ++ library Version: 19.4.0 Libs: -L $ {libdir} -ldlib Cflags: -I $ {includedir} Requires: libpng

As I need to report the progress to the teacher ,looking forward to your advice, thank you again.

At 2018-01-04 09:54:46, "Jesmond" [email protected] wrote:

@ml930310 Put seqmap file, which stores the relative path (from data/) to your detection data, under data/. The content of the seqmap is something like this:

../../dataset/MOT16/train/MOT16-02 ../../dataset/MOT16/train/MOT16-04 ../../dataset/MOT16/train/MOT16-05 ../../dataset/MOT16/train/MOT16-09 ../../dataset/MOT16/train/MOT16-10 ../../dataset/MOT16/train/MOT16-11 ../../dataset/MOT16/train/MOT16-13

Then run ./trackApp.out -s -m -f , example ./trackApp.out -s seqmap.txt -m det -f mot. It will locate these folders and perform tracking using the det.txt with the path mentioned, e.g. ../../dataset/MOT16/train/MOT16-13/det/det.txt. I have only tried the tracking part, so I do not know much about setting the detection model. I guess you can try to read the demo codes under examples/ for more information.

But my way of doing this might be wrong. The values and results i got from the tracking are quite weird #5. Do you get this kind of result too?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

ml930310 avatar Jan 09 '18 11:01 ml930310

@ml930310 I think your caffe is not installed properly. Check your installation again and try to run a simple network first

jazzseow avatar Jan 11 '18 04:01 jazzseow

@ml930310 It is easier run demo by make compile and make track. It will build all the necessary links between the libraries and the target file for you.

Hello , I have some mistake to do "make track",the message is : (link) CXX trackApp.out build/tracker/predictor/kalman/KalmanPredictor.o:在函数‘dlib::lapack::binding::gesvd(char, char, int, int, double*, int, double*, double*, int, double*, int, double*, int)’中: /usr/local/include/dlib/matrix/lapack/gesvd.h:38:对‘dgesvd_’未定义的引用 collect2: error: ld returned 1 exit status makefile:84: recipe for target 'trackApp.out' failed make: *** [trackApp.out] Error 1

I have try to deal with it , but failed , so can you give me some advise? thank you very much.

zhangmengzxz avatar Nov 05 '18 03:11 zhangmengzxz