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

Purpose of pose estimations in tod_detecting?

asked 2011-03-23 02:48:18 -0500

Julius gravatar image

What is the purpose of the pose estimations in tod_detecting when recognizing textured objects using "rosrun tod_detecting recognizer --mode=0 ..."? Are they relevant for the recognition process (not running kinect mode)?

image description

(picture showing the estimated pose of a training object, as described by rvec and tvec)

Why do we need to make pose estimations in the tod_training pipeline, apart from using the pose estimation to segment the point cloud to generate a ROI mask? I mean the training pipeline does not really collect any real 3d features.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
1

answered 2011-03-23 07:59:59 -0500

updated 2011-03-23 08:30:04 -0500

I don't understand first part of your question. We do pose estimation in training for calculation transformation between frames.

mode = xx

Now tod_detecting contains two recognizers. First version of TOD (textured_object_detection package) was implemented for recognizing objects on 5 MP prosilica dataset. Later Ethan and me began to refactor this code. In result we get two versions of detection algorithms: Kinect mode and TOD mode. Kinect mode gives better results on Kinect data, TOD mode gives better results on 5MP data.

edit flag offensive delete link more

Comments

Thanks! I just wondered where the pose estimates are used in detecting objects. In other words, is the pose estimation required in tod_detecting and if so, why? See http://answers.ros.org/question/537/are-my-findings-about-tod_detecting-correct
Julius gravatar image Julius  ( 2011-03-25 06:25:55 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2011-03-23 02:48:18 -0500

Seen: 285 times

Last updated: Mar 23 '11