sound_corr_temp == np.nanmax(sound_corr_temp))[0][0]
else:
mvmt_corr_temp = copy.deepcopy(mvmt_corr)
mvmt_corr_temp[sound_channel] = 0
mvmt_channel = np.where(
mvmt_corr_temp == np.nanmax(mvmt_corr_temp))[0][0]
average = np.mean(np.vstack([mvmt_corr, sound_corr]), axis=0)
rest_channel = np.where(average == np.nanmin(average))[0][0]
# plt.plot(sound_sections, label="actual sound")
# plt.plot(mvmt_sections, label="actual mvmt")
# plt.plot(cluster_sections[:,mvmt_channel], label="cluster mvmt")
# plt.plot(cluster_sections[:,sound_channel], label="cluster sound")
# plt.plot(cluster_sections[:,rest_channel], label="cluster rest")
# plt.legend()
# plt.show()
#
# pdb.set_trace()
print str(mvmt_channel) + str(mvmt_corr[mvmt_channel])
print str(sound_channel) + str(sound_corr[sound_channel])
print str(rest_channel) + str(average[rest_channel])
return {'Mvmt': mvmt_channel, 'Sound': sound_channel, 'Rest': rest_channel}
if __name__ == "__main__":
if not (len(sys.argv) == 4):
raise varError("Arguments should be <Sound File> <Movement File>\
<Cluster Result File>")
correlate(*sys.argv[1:])
else:
cluster_centers.append([])
pickle.dump(cluster_centers, open(save_loc + "\\" + sbj_id + "_" +
str(date) + "_" + str(level) + "_centers.p", "wb"))
# Plot Cluster Figures
plt.figure(figsize=(20,10))
for l in xrange(5):
condensed = condense_time(cluster_result[l, :], l, 16)
index = np.argsort(condensed[:,:].sum(axis=0))
# for c in index[-min(2**(l+1),5):]:
# plt.plot(np.arange(0,condensed.shape[0]*10,10),condensed[:,c])
#
# plt.xlabel("Time(Minute)")
# plt.ylabel("Cluster Count")
# #plt.xticks(np.arange(0,condensed.shape[0]*10,10))
# plt.show()
pickle.dump(condensed, open(save_loc + "\\" + sbj_id + "_" + str(date) + "_" + str(l) + ".p", "wb"))
pickle.dump(cluster_result, open(save_loc + "\\" + sbj_id + "_" + str(date) + ".p", "wb"))
if __name__ == "__main__":
if not (len(sys.argv) == 5):
raise varError("Arguments should be <Subject ID> <Dates> <Features directory>\
<Save directory>")
sys.argv[2] = sys.argv[2].split()
hier_cluster_main(*sys.argv[1:])
"Speaking", "Multiple_people", "Sleeping", "Eating",
"Listening.Watching_Media", "Listening.Listening_to_family_member",
"Listening.Listening_to_staff", "Rest"
]
tracks_reduced = ["Mvmt", "Sound", "Rest", "Other"]
for i in xrange(800):
file_num = str(i).zfill(4)
sbj_indexes = np.where(
np.array(labels.filename) == file_num + '-' + labeller)[0]
if sbj_indexes.shape[0] > 0:
vid_length = getLength(vid_folder + "\\" + sbj_id + "_" +
str(day) + "\\" + sbj_id + "_" + str(day) +
"_" + file_num + ".avi")
result = convert_reduced_labels_to_array(labels, sbj_indexes,
vid_length, tracks,
tracks_reduced)
pickle.dump(
result,
open(
dst_folder + "\\" + sbj_id + "_" + str(day) + "_" +
file_num + "_" + labeller + ".p", "wb"))
if __name__ == "__main__":
if not (len(sys.argv) == 7):
raise varError("Arguments should be <Subject ID> <Day><Label File>\
<Video directory> <Save directory>")
extract_reduced_labels(*sys.argv[1:])
sound_corr_temp[mvmt_channel]=0
sound_channel = np.where(sound_corr_temp==np.nanmax(sound_corr_temp))[0][0]
else:
mvmt_corr_temp = copy.deepcopy(mvmt_corr)
mvmt_corr_temp[sound_channel]=0
mvmt_channel = np.where(mvmt_corr_temp==np.nanmax(mvmt_corr_temp))[0][0]
average = np.mean(np.vstack([mvmt_corr, sound_corr]), axis=0)
rest_channel = np.where(average==np.nanmin(average))[0][0]
# plt.plot(sound_sections, label="actual sound")
# plt.plot(mvmt_sections, label="actual mvmt")
# plt.plot(cluster_sections[:,mvmt_channel], label="cluster mvmt")
# plt.plot(cluster_sections[:,sound_channel], label="cluster sound")
# plt.plot(cluster_sections[:,rest_channel], label="cluster rest")
# plt.legend()
# plt.show()
#
# pdb.set_trace()
print str(mvmt_channel) + str(mvmt_corr[mvmt_channel])
print str(sound_channel) + str(sound_corr[sound_channel])
print str(rest_channel) + str(average[rest_channel])
return {'Mvmt':mvmt_channel,'Sound':
sound_channel,'Rest':rest_channel}
if __name__ == "__main__":
if not(len(sys.argv) == 4):
raise varError("Arguments should be <Sound File> <Movement File>\
<Cluster Result File>")
correlate(*sys.argv[1:])
def extract_labeller_reduced_labels(sbj_id, day, src_folder, vid_folder, dst_folder, labeller):
labels=pd.read_csv(src_folder + sbj_id + "_" + str(day) + ".txt", sep=':', dtype=str)
tracks = ["Laughing", "Movement.Head", "Movement.Other",
"Movement.arm", "Speaking", "Multiple_people",
"Sleeping","Eating", "Listening.Watching_Media",
"Listening.Listening_to_family_member",
"Listening.Listening_to_staff", "Rest"]
tracks_reduced = ["Mvmt", "Sound", "Rest", "Other"]
for i in xrange(800):
file_num = str(i).zfill(4)
sbj_indexes = np.where(np.array(labels.filename)== file_num + '-' + labeller)[0]
if sbj_indexes.shape[0]>0:
vid_length = getLength(vid_folder + "\\" + sbj_id
+ "_" + str(day) + "\\" + sbj_id
+ "_" + str(day) + "_" + file_num + ".avi")
result = convert_reduced_labels_to_array(labels,sbj_indexes, vid_length, tracks, tracks_reduced)
pickle.dump(result, open(dst_folder + "\\" + sbj_id +"_" +str(day) +
"_" + file_num + "_" + labeller + ".p", "wb"))
if __name__ == "__main__":
if not(len(sys.argv) == 7):
raise varError("Arguments should be <Subject ID> <Day><Label File>\
<Video directory> <Save directory>")
extract_reduced_labels(*sys.argv[1:])
