ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Problem with face recognition using kinect

asked 2014-04-30 23:41:04 -0500

smart engineer gravatar image

updated 2015-09-01 00:04:18 -0500

130s gravatar image

Hello ; I am trying to make a face Recognition using the following package """ https://github.com/procrob/procrob_fu. . . but unfortunately I face many Challenges : 1_ this package provides the method of face recognition with webcam but in my project i use kinect ... 2_ when I apply the package on webcam images don't be added to the data file nor the train.txt 3_i can't understand what meant by "order_id" and how can i use it? i am working on ubuntu 12.04 and fuerte distribution ...

edit retag flag offensive close merge delete

Comments

The link seems to be broken. Please update it.

karthik_ms gravatar image karthik_ms  ( 2014-05-01 00:29:56 -0500 )edit
1

I have used this package with groovy and a kinect and it works well if you look at the read me file the order_id is the command to add images and retrain the database

Peter Heim gravatar image Peter Heim  ( 2014-05-01 12:21:46 -0500 )edit

Thanx for your help ; I execute the following commands :: 1_roscore 2_roslaunch openni_launch openni.launch 3-rosrun rviz rviz 4-rosrun rqt_reconfigure rqt_reconfigure 5_rosrun image_view image_view image:=/camera/rgb/image_color 6-roscd face_recognition 7-rosrun face_recognition Fserver 8-rosrun face_recognition Fclient till here every thing work , but the rostopic part ,i can't understand how to use it at all . """rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 2 "your_name",,,, rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 1 "none" ,,,, rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 2 "your_friend's_name",,, rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 0 "none",,,, rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 1 "none" ,,,, rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 4 "none" ,,,,"""" Is there any differences will occur if I use Groovy or Fuerte distribution ????

smart engineer gravatar image smart engineer  ( 2014-05-02 00:57:13 -0500 )edit

Can you edit your question to add the commands you're running, and maybe spend some time trying to format it so that it's readable?

ahendrix gravatar image ahendrix  ( 2014-05-02 06:16:01 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted
0

answered 2014-05-02 21:39:41 -0500

Peter Heim gravatar image

You have to run the programs from the bin folder also you need to move the other files into that folder from memory the data folder need to be below the bin folder. If you search for face recognition it will have the details make sure you publish the camera in the correct topic. You may need to train the database first to train the database run rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 3 "none" there is no reason to run image_view when you run rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 0 "none" a window will open showing the recognition results

edit flag offensive delete link more

Comments

@Peter Heim ; How can I run the program from the bin folder ? what files must be moved and how?

smart engineer gravatar image smart engineer  ( 2014-05-03 10:52:18 -0500 )edit

cd to face_recognition/bin then rosrun face_recognition Fserver in another terminal rosrun face_recognition Fclient transfer train.txt, haarcascade_frontalface_alt.xml and the data folder haarcascade_frontalface_alt

Peter Heim gravatar image Peter Heim  ( 2014-05-03 19:53:09 -0500 )edit

@Peter Heim ::: I transfer the following files to the Bin folder """ data folder ,, train.txt ,,, haarcascade_frontalface_alt.xml ,,,facedata..xml """ Then I Execute the following commands ::: 1_ roscore 2_roslaunch openni_launch openni.launch 3_rosrun rviz rviz >>> and here i edit the fixed frame to ""camera_depth_optical_frame""and added PointCloud2 and set the topic to /camera/depth/points 4_rosrun image_view image_view image:=/camera/rgb/image_color 5_roscd face_recognition 6_rosrun face_recognition Fserver 7_rosrun face_recognition Fclient 8_rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 2 "Somaya" 9_rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 0 "none" 10-rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 1 "none" 11_rostopic pub -1 /fr_order face_recognition/FRClientGoal -- 4 "none" But still no saved images ,, what is wrong ???? and what the variable "none" in the rostopic pub-1 refers to ????

smart engineer gravatar image smart engineer  ( 2014-05-04 07:47:01 -0500 )edit

make sure the kinect is publishing images to /camera/image_raw you can use rxgraph to see if they are connected. I found it easier to change the topic in face_recognition.ccp and redo rosmake. the none in the other 3 topics is just a place holder. when the server starts what does it print out

Peter Heim gravatar image Peter Heim  ( 2014-05-04 11:59:40 -0500 )edit

@Peter Heim I changed the face_recognition.cpp """" //subscribe to video stream through image transport class image_sub_ = it_.subscribe("/camera/image_raw", 1, &FaceRecognition::imageCB, this) """" after that when I tried to run the ""rosrun face_recognition Fserver"" it printed [ INFO] [1399282787.174585362]: Training data loaded (50 training images of 2 people): [ INFO] [1399282787.174876259]: People: [ INFO] [1399282787.174944313]: <sergio> [ INFO] [1399282787.175038382]: , <pouyan> terminate called after throwing an instance of 'ros::InvalidNameException' what(): Character [ ] is not valid as the first character in Graph Resource Name [ /camera/image_raw]. Valid characters are a-z, A-Z, / and in some cases ~. Aborted (core dumped .... Before Editing the face_recognition.cpp>>> it printed the following "" [ INFO] [1399282787.174585362]: Training data loaded (50 training images of 2 people): [ INFO] [1399282787.174876259]: People: [ INFO] [1399282787.174944313]: <sergio>

smart engineer gravatar image smart engineer  ( 2014-05-04 23:45:14 -0500 )edit

You have to change it to the topic published by your kinect don't use cut and past type it in i have found that will cause errors. Using rxgraph the face recognition node must connect to the kinect topic or no images will be produced

Peter Heim gravatar image Peter Heim  ( 2014-05-10 22:02:30 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2014-04-30 23:41:04 -0500

Seen: 577 times

Last updated: May 02 '14