ros_tensorflow

Introduction

This repo introduces how to integrate Tensorflow framework into ROS with object detection API.

And through this repo, you can realize mnist, object recognition, and object detection respectively.

Requirements

• Ubuntu 16.04 with Python2.7
• Install ROS(Kinetic) with catkin build. Create a catkin workspace.
• Install Tensorflow(1.2.0-1.10.0 all be ok)
• Install CUDA 6.5 for NVIDIA TK1 (Choosing according to your needs)
• Some dependencies

  sudo apt-get install protobuf-compiler python-pil python-lxml
  sudo pip install jupyter
  sudo pip install matplotlib
  

Grab the source

cd [CATKIN_WS]/src
git clone https://github.com/cong/ros_tensorflow.git

Build

Build ROS package by

cd [CATKIN_WS]
catkin_make

PS: Python doesn't seem to need to be compiled.

Run

For mnist:

# First, open a terminal, execute
roscore

# Second, Open another terminal, then execute
# Please install "usb_cam" node before you execute blow
roslaunch usb_cam usb_cam-test.launch

# Third, open another terminal, then execute
roslaunch ros_tensorflow ros_tensorflow_mnist.launch
# You can echo a topic to receive the string message.
rostopic echo /result_ripe

For object recognition:

# Third, open another terminal, then execute
roslaunch ros_tensorflow ros_tensorflow_classify.launch
# You can echo a topic to receive the string message.
rostopic echo /result_ripe

For object detection:

# Third, open another terminal, then execute
roslaunch ros_tensorflow ros_tensorflow_detect.launch
# You can through "image_view" node to receive images detected.
rosrun image_view image_view image:=/result_ripe

ROS Topics

Publish a topic : /result_ripe

Receive an image : usb_cam/image_raw

Optional setting

• You can realize your project by replacing the files in "ros_tensorflow/include/" with your own files.
• If you feel that it is helpful to you, please give me a star. Thx! :)
• For more information you can visit the Blog.