This is the tensorflow slim model for emotion reconition
Ret1): 表情代號: 0:angry ,1:happy ,2:neutral ,3:sad ,4:surprise
Ret2): 情緒百分比: %
Ret3): 臉部坐標: (x, y, w, h) == (46, 202, 127, 127)
$ (表情,%百分比,臉部坐標[x,y,w,h] ) == (Happy,0.8253038,[46,202,127,127])- Install TensorFlow (see tensor flow install guide)
- Install ROS (see http://wiki.ros.org)
- InstallROSTX2 (see https://github.com/jetsonhacks/installROSTX2)
- updateRepositories.sh
- installROS.sh
- setupCatkinWorkspace.sh
- Install cv-bridge
$ sudo apt-get install ros-kinetic-cv-bridge ros-kinetic-opencv3- (Optional1) Install camera driver (for example, cv_camera , cv_camera_node)
$ sudo apt-get install ros-kinetic-cv-camera- (Optional2) Install UVC Camera for ROS
$ sudo apt-get install ros-kinetic-libuvc-camera
$ sudo apt-get install ros-kinetic-image-pipeline
$ rosdep update- (Optional3) Create udev rule
- /etc/udev/rules.d/99-uvc.rules
$ SUBSYSTEM=="usb", ATTR{idVendor}=="046d", MODE="0666"- (Optional4) Test Camera
$ rosrun libuvc_camera camera_node
$ rosrun image_view image_view image:=image_rawPlease read official guide. This is a only note for me.
$ sudo apt-get install python-pip python-dev python-virtualenv
$ virtualenv --system-site-packages ~/tensorflow
$ source ~/tensorflow/bin/activate
$ pip install --upgrade tensorflowPlease read official guide. This is a only note for me.
$ sudo apt-get install python-pip python-dev python-virtualenv
$ virtualenv --system-site-packages ~/tensorflow
$ source ~/tensorflow/bin/activate
$ pip install --upgrade tensorflow-1.4.1-cp27-cp27mu-linux_aarch64.whl
$ https://github.com/peterlee0127/tensorflow-nvJetson
$ https://github.com/jetsonhacks/installTensorFlowTX2- subscribe: /CVsub (sensor_msgs/Image)
- publish: /CVpub (sensor_msgs/Image)
- publish1: /result (std_msgs/String)
Opencv 3.2 or laster
Tensorflow 1.4.1
train_image_size=160
max_number_of_steps=60000
Pre-trained Models=resnet_v2_50
How to try
old version
***
$ source /root/catkin_ws/devel/setup.bash
$ roscore
$ rosrun libuvc_camera camera_node
$ rosrun image_view image_view image:=image_raw
$ rosrun image_view image_view image:=CVpub
$ rostopic echo /result
$ rosrun rqt_graph rqt_graph
$ source ~/tensorflow/bin/activate
$ python main1.py CVsub:=image_rawnew version
***
$ cd /root/catkin_ws/
$ catkin_make
$ source /root/catkin_ws/devel/setup.bash
$ roscore
$ roslaunch usb_cam usb_cam-test.launch
$ rosrun rosEmotion_detect listen.py CVsub1:=/usb_cam/image_raw
cat result.log
See main1.py
#Program execution
python main1.py CVsub:=image_raw
#function output
(表情,%百分比,臉部坐標[x,y,w,h] ) == (Happy,0.8253038,[46,202,127,127])
See main.py
#function input
expr_list = r.classify_image(frame, bbox_list)
#function output
#[(2, 0.93376148), (2, 0.98358649), (1, 0.98518467), (2, 0.9335444), (0, 0.7758944)]
print(fexpr_list)
