An on going TF implementation on SqueezeSeg to deal with LiDAR pointcloud.
bin2depth.py: is used to generate depth map and the corresponding ground truth from kitti-velodyne-binary file, by calibration velo2cam.
squeeze_seg.py: is the main part of the network(still under construction, the current part only finish the CNN part but the CRF-AS-RNN refinement).
pcm2pcm.py: is a ROS node which listens pointcloud2 message from ROS, and republish pointcloud2 message with squeeze_seg predicted labels. Both of the pc2 message topic to be listenned and the pc2 message topic to be published can be defined in the cmd.
