ros_opencl_caffe

Introduction

OpenCL Caffe(clCaffe) is an OpenCL backend for Caffe from Intel®. With inference optimized by Intel® OpenCL, clCaffe can be used in most scene in high performance, like objects inference.

This project is a ROS wrapper for OpenCL Caffe, providing following features:

A ROS service for objects inference in a ROS image message
A ROS publisher for objects inference in a ROS image message from a RGB camera
Demo applications to show the capablilities of ROS service and publisher

Prerequisite

An x86_64 computer with Ubuntu 16.04
ROS Kinetic
RGB camera, e.g. Intel® RealSense™, Microsoft® Kinect™ or standard USB camera

Environment Setup

Install ROS Kinetic Desktop-Full (guide)
Create a catkin workspace (guide)
Install clCaffe (guide)
Create a symbol link in /opt/clCaffe

  sudo ln -s <clCaffe-path> /opt/clCaffe

Add clCaffe libraries to LD_LIBRARY_PATH.

  export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/clCaffe/build/lib

Install ROS package for different cameras as needed. e.g.
1. Standard USB camera
```
  sudo apt-get install ros-kinetic-usb-cam
```
1. Intel® RealSense™ camera
- Install Intel® RealSense™ SDK 2.0 (guide). Refer here for more details about Intel® RealSense™ SDK 2.0.
- Install Intel® RealSense™ ROS (guide)
```
  cd ~/catkin_ws/src
  git clone https://github.com/intel-ros/realsense.git
  cd  realsense
  git checkout 2.0.0
  cd ~/catkin_ws
  catkin_make
```
1. Microsoft® Kinect™ camera
```
  sudo apt-get install ros-kinetic-openni-launch
```

Building and Installation

  cd ~/catkin_ws/src
  git clone https://github.com/intel/object_msgs
  git clone https://github.com/intel/ros_opencl_caffe
  cd ~/catkin_ws/
  catkin_make
  catkin_make install
  source install/setup.bash

Copy object label file to clCaffe installation location

  cp ~/catkin_ws/src/ros_opencl_caffe/opencl_caffe/resources/voc.txt /opt/clCaffe/data/yolo/

Running the Demo

Inference

Standard USB camera

    roslaunch opencl_caffe_launch usb_cam_viewer.launch

Intel® RealSense™ camera

    roslaunch opencl_caffe_launch realsense_viewer.launch

Microsoft® Kinect™ camera

    roslaunch opencl_caffe_launch kinect_viewer.launch

Service

  roslaunch opencl_caffe_launch opencl_caffe_srv.launch

Test

Use rostest for tests

  source devel/setup.bash
  rostest opencl_caffe service.test
  rostest opencl_caffe detector.test

Known Issues

Only image messages supported in service demo
Only test on RealSense D400 series camera, Microsoft Kinect v1 camera and Microsoft HD-300 USB camera

TODO

Support latest clCaffe

MengNan-Li / ros_opencl_caffe