-
Notifications
You must be signed in to change notification settings - Fork 1
Home
This ros package is to show the SRNN prediction about some human activity useing the CAD120 dataset.
If you want to get more information about SRNN and dataset CAD120,please access :
ubuntu = 16.04
Ros = kinetic
theano = 0.9
numpy = 1.9.3
matplotlib = 2.0.2
nose = 1.1.2
You can access this website to get video:
Youtube: https://www.youtube.com/channel/UCVpZP1oKUq2jH4agHBC0oMQ
Youku: https://i.youku.com/i/UNTYyMTc4ODU5Mg==/videos?spm=a2hzp.8244740.0.0
Beofore you run this node, please make sure your computer have been download theCAD120 dateset you can get from this webist http://pr.cs.cornell.edu/humanactivities/ and put it in the .src/activity-anticipation/src/ .
mkdir -p SRNN/src && cd SRNN/src/
git clone https://github.com/Jiongyu/activity-anticipation
cd ..
catkin_make
source devel/setup.bash
Firstly,you should to run the readData.py to generate train,test,groud truth file.
rosrun activity-anticipation readData.py
Secondly, you should to run the activity-rnn-full-model.py to train model and it will genearte checkpoints file.
rosrun activity-anticipation activity-rnn-full-model.py
Thirdly,you should to run the activity_prediction_sharedrnn.py to generate the prediction file.
rosrun activity-anticipation activity_prediction_sharedrnn.py
Finally,you should to run the show-prediciton.py and RVIZ to show the consquence about detection and anticipation.
rosrun activity-anticipation show-prediciton.py
All right , you will see the consquence in the RVIZ.
- how to you get the mean of the pik file about ground truth,test,train,prediction?