matlabbe

Make sure those topics are published: ```bash $ rostopic hz /odom /camera/rgb/image_raw /camera/depth/image_raw /camera/rgb/camera_info /scan subscribed to [/odom] subscribed to [/camera/rgb/image_raw] subscribed to [/camera/depth/image_raw] subscribed to [/camera/rgb/camera_info] subscribed to [/scan]...

Set `subscribe_scan` to false.

It could be indeed nice to have, but I would like to use it more like a library to be less difficult to maintain. The project doesn't export the functions...

First draft here (CUDA not tested): https://github.com/introlab/find-object/commit/b07239a0003e67dba1ecef5eff126f2f352eba7f

On my system (with OpenCV3), I don't have a problem with IplImage (which is defined in /usr/local/include/opencv2/core/type_c.h). However, it is not used anymore, cv:Mat is used everywhere so I removed...

It may be nice to have an option like "Edit->Add descriptors to vocabulary (*.txt)" to push back descriptors saved in a text file to the vocabulary.

If you refer to 3D example, it is computed using the depth image. It takes depth pixels in the middle of the object detected.

You should have to convert the rgb image. I updated the launch example: https://github.com/introlab/find-object/commit/a1601f7a24ec416264c6ddb6c7cb47a8c91c1cea The code used to find the 3d pose from the depth image works only well with...

find-object is very features2d oriented, as it was the goal at the beginning. For blob detector, I think there are some examples to use out of the box, even in...

Hi Nicolas, By maintaining objects in RVIZ, do you mean that when the robot is not seeing anymore a detected object, the position of the object can be still visualized...