Np average

[Charlie839242]

In the file https://github.com/Charlie839242/examples/blob/master/lite/examples/pose_estimation/raspberry_pi/data.py, 111 line, it is possible that the score of 17 keypoints all don't reach the threshold. In this case, the np.average receives a empty list, which will cause some warnings like: RuntimeWarning: Mean of empty slice. RuntimeWarning: invalid value encountered in double_scalars

By adding 'scores_above_threshold = np.append(scores_above_threshold, 0)' can solve the problem.