Introduction

Add your content here.

Requirements

Ubuntu 12.04 or above with ROS Groovy

Dependencies

• cv_image_source : ROST uses cv_image_source ROS package by default for getting images from camera or video files. To install it:
  • $ hg clone https://bitbucket.org/yogeshgirdhar/cv_image_source
  • $ rosmake cv_image_source
• system dependencies :
  • $ sudo apt-get install libctemplate-dev libsndfile1-dev libyaml-cpp0.3-dev libpulse-dev libfftw3-dev

Building ROST

• $ hg clone https://code.google.com/p/rost-ros/
• $ rosmake rost-ros

Using ROST

Using ROST with camera

$ roslaunch rost_launch rost.launch camera:=true [options]

Using ROST with a video file

$ roslaunch rost_launch rost.launch file:=/absolute/path/of/video.mp4 [options]

Optional arguments the the launch file(with default values shown): * alpha:=0.1 controls the smoothness of topic distribution describing an image patch * beta:=1.0 controls the smoothness of word distribution describing a topic * tau:=2.0 how important is present compared to past for topic refinement. 1. 0 => all times are equally important

tau > 1 => present is more important 0 past is more important * K:=16 number of topics * S:=16 summary size * vout:= record the visualization video. output is MJPEG avi file. * video.rate:=5 process video at 5 fps * video.subsample:=2 temporally subsample the video (only applicable when using file.launch) * video.loop:=false loop the video (only with file.launch) * video.scale:=1.0 scale the input video. To process the video at half the linear size, set scale:=0.5 * threads:=4 number of threads to spawn for topic refinement. * cell.width:=160 split the input image into windows of size 160x160 * image.width:=640 tell the system the size of the input vide * image.height:=480

There are many more options. To see them open the launch file.