add data to training set

The provided train data may not sufficient to get good pose recognization results. In case the pose is not recognized correctly, this data can be manually added to the train data.

Defined Poses

Stand: the weight is supported by the feet and the torso is upright.
StandInit: the torso is upright, and legs are bended (be ready to walk)
Sitting: the buttock is in contact with the ground and the torso is upright.
Crouch: sit on its feet
Belly: fall, stretched and facing down
Back: fall, stretched and facing up
Forg: fall, facing down with the trunk lifted.
HeadBack: fall, facing up with the trunk lifted.
Left: fall and facing to its left side
Right: fall and facing to its right side

There are images in robot_pose_image folder as examples for predefined poses.



In [7]:

    
%pylab inline
imshow(imread('robot_pose_image/Stand.png'))









    



Populating the interactive namespace from numpy and matplotlib






    Out[7]:





<matplotlib.image.AxesImage at 0x7fa1552dca90>

Load avaiable data



In [8]:

    
import pickle
from os import path

pose_name = 'Back'
data_file = path.join('robot_pose_data', pose_name)
data = pickle.load(open(data_file))



In [9]:

    
print data









    



[[-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7180380821228027, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8945322036743164, -0.008992496877908707, -1.516520619392395], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7180380821228027, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8960661888122559, -0.006883446592837572, -1.5203558206558228], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7165040969848633, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8945322036743164, -0.006883446592837572, -1.520164132118225], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7165040969848633, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8945322036743164, -0.007075135130435228, -1.5199724435806274], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7165040969848633, -0.05058002471923828, -0.1487560272216797, -0.9296460151672363, 1.8945322036743164, -0.007266823668032885, -1.5199724435806274], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7180380821228027, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8945322036743164, -0.007266823668032885, -1.5197802782058716], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7165040969848633, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8945322036743164, -0.006883446592837572, -1.5216976404190063], [-0.05058002471923828, -0.22852396965026855, -0.8298521041870117, 1.7165040969848633, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8960661888122559, -0.008992496877908707, -1.5134526491165161], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7165040969848633, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8960661888122559, -0.008992496877908707, -1.5161372423171997], [-0.05058002471923828, -0.2269899845123291, -0.8298521041870117, 1.7180380821228027, -0.05058002471923828, -0.1487560272216797, -0.9311800003051758, 1.8945322036743164, -0.009951416403055191, -1.5161372423171997], [0, 0, 0, 0, 0, 0, 0, 0, 0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966], [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, -1.5707963267948966]]

Add new data



In [10]:

    
new_data = [0, 0, 0, 0, 0, 0, 0, 0, 0, -1.5707963267948966]
data.append(new_data)
pickle.dump(data, open(data_file, 'w'))

Now, you have to retrain your classifier learn_posture.ipynb again.



In [ ]: