Ask Your Question

Region_tlr_ssd doesn't preview on rviz (ImageViewerPlugin]) and no signal

asked 2019-05-06 22:10:41 -0500

Eng-Mo gravatar image

updated 2019-05-07 01:26:00 -0500

Hi guys,

I am trying to turn on the traffic light recognition using region_tlr_ssd but it doesn’t preview any result on rviz. I followed the ssdcaffe installation from [ ] and installation and compilation done without any error.

I used the pretrained model that mentioned in the docs and save it in: Autoware/ros/src/computing/perception/detection/trafficlight_recognizer and the model has loaded but I get the following message:

Ignoring source layer data
Ignoring source layer data_data_0_split
Ignoring source layer mbox_loss

The steps as follow:

  1. initialized localization by run the TF, vehicle model from setup
  2. run point cloud , vector map, and TF
  3. launched velodyne from sensing and turn on the camera mindvision, launched the camera calibration.
  4. launched ndt_matching, feat_proj, and region_tlr_ssd
  5. opened the rviz and added new panel and chose tlr_superimpose topic.

System Specification

  1. Ubuntu 16.4
  2. Autoware 1.7
  3. Cuda 9
  4. ROS Kintenic

Thank you

edit retag flag offensive close merge delete

1 Answer

Sort by » oldest newest most voted

answered 2019-05-28 23:25:00 -0500

Eng-Mo gravatar image

Hi Guys,

I Mostly know the problem, the problem in the victor map. the vector map has to be very accurate otherwise it will never work at most cases, and the one which it used in the example is done by professional company and I am using a real car in real environment and don't have a ccurate vector map that includes all the road component.


edit flag offensive delete link more

Your Answer

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

Add Answer

Question Tools

1 follower


Asked: 2019-05-06 22:06:27 -0500

Seen: 117 times

Last updated: May 28 '19