ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | Q&A
Ask Your Question

object detection

asked 2014-07-14 05:49:46 -0500

Massbuilder gravatar image

I have a bit of an interesting question for you. I have been experimenting with Opencv for a while and would like to add so vision applications to my robot, especially object detection. However I cannot to get either template matching or matching via key points to work. What methods would you suggest?

edit retag flag offensive close merge delete


It highly depends on your performance/speed constraints. From my own experience, using keypoints works quite well (SIFT and SURF being the most accurate but slowest). Are you doing keypoints detection, descriptors extraction and keypoints matching with outlier removal (using RANSAC)?

Mehdi. gravatar image Mehdi.  ( 2014-07-15 04:48:46 -0500 )edit

I can only comment that I had satisfying results using HOG features in combination with a SVM classifier.

Hansg91 gravatar image Hansg91  ( 2014-07-15 07:34:59 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2014-07-15 04:29:31 -0500

DevonW gravatar image

updated 2014-07-15 09:33:23 -0500

There's no magic solution my friend. You have to stick through it and see it to the end

To answer the second part:

There's deep neural networks that do object detection. Pylearn2 implements a lot of these deep neural nets with easy to use config files. There are tutorials on pylearn2:

I have code implemented that does a HoG SVM 2 layer then classify situation. It doesn't work very well, but it's a start. It's also in ROS, but has not been implemented as a ROS node yet. You can look through this to see how the keypoints work for your above situation.

Here is how the keypoints are extracted:

edit flag offensive delete link more



well at least it is a start. I will be sure to investigate these leads.

Massbuilder gravatar image Massbuilder  ( 2014-07-15 17:08:47 -0500 )edit

You should also check the object detection algorithm from Roboearth, the advantage is that it is cloudbased and many object are already trained. For creating an object model you need a Kinect but for recognition tasks it is possible to do it with a monocular camera.

Mehdi. gravatar image Mehdi.  ( 2014-07-15 18:56:15 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools



Asked: 2014-07-14 05:49:46 -0500

Seen: 487 times

Last updated: Jul 15 '14