def compare(k):
N_INDCS = -1
map_test, map_train = map_overlay.basic_setup([100, 400], 50, label_name = "Jared")
clf = ObjectClassifier()
damage_labels = np.loadtxt('damagelabels50/Jared-3-3.csv', delimiter = ',').astype('int')
building_labels = np.loadtxt('damagelabels50/all_rooftops-3-3.csv', delimiter =',').astype('int')
segs = map_train.segmentations[50].ravel()
n_segs = int(np.max(segs))+1
all_labels = np.zeros(n_segs) #Nothing = Green
all_labels[damage_labels] = 1 #Damage = Red
all_labels[building_labels] = 2 #Building = Blue
all_data, y = clf._get_X_y(map_train, 'Jared')
clf.fit(map_train, label_name = 'Jared') #<-this one changesthe training for the importance order
names = clf.feat_names
print names
importances = clf.model.feature_importances_
indices = np.argsort(importances)[::-1]
print names[indices][:N_INDCS]
plt.figure()
std = np.std([tree.feature_importances_ for tree in clf.model.estimators_],
axis=0)
plt.title("Feature importances")
plt.bar(range(names.shape[0]), importances[indices],
color="r", yerr=std[indices], align="center")
plt.xticks(range(names.shape[0]), names[indices])
plt.xlim([-1, names.shape[0]])
all_histos(all_data[damage_labels], all_data[building_labels], all_data[np.where(all_labels == 0)[0]], names, indices[:N_INDCS])
#plt.show()
return names, indices
'''
def comparepx(k):
N_INDCS = -1
map_test, map_train = map_overlay.basic_setup([100], 50, label_name = "all_buildings")
segs = map_train.segmentations[50].ravel().astype('int')
clf = PxClassifier(85,-1)
damage_labels = np.loadtxt('damagelabels50/Jared-3-3.csv', delimiter = ',').astype('int')
building_labels = np.loadtxt('damagelabels50/all_buildings-3-3.csv', delimiter =',').astype('int')
n_segs = int(np.max(segs))+1
all_labels = np.zeros(n_segs) #Nothing = Green
all_labels[damage_labels] = 1 #Damage = Red
all_labels[building_labels] = 2 #Building = Blue
all_labels = all_labels[segs].ravel()
#plt.imshow(all_labels.reshape(map_train.rows, map_train.cols), cmap = 'gray')
indcs = np.random.choice(np.arange(map_train.rows*map_train.cols), 1000000, replace = False).astype('int')
#print np.where(all_labels == 0)[0][indcs].shape
#print np.where(all_labels == 1)[0][indcs].shape
#print np.where(all_labels == 2)[0][indcs].shape
all_data, names = clf.gen_features(map_train)
#print all_data[:,:3].shape, map_train.rows, map_train.cols
#plt.imshow(all_data[:,:3].reshape(map_train.rows, map_train.cols, 3))
#plt.show()
#print all_data[np.where(all_labels == 1)[0]][indcs], all_data[np.where(all_labels[indcs] == 1)[0]].shape
'''clf.fit(map_train)
clf.feature_importance()
plt.show()
names = clf.feat_names
importances = clf.model.feature_importances_
indices = np.argsort(importances)[::-1]'''
#print names[indices][:N_INDCS]
all_histos(all_data[indcs][np.where(all_labels[indcs] == 1)[0]], all_data[indcs][np.where(all_labels[indcs] == 2)[0]], all_data[indcs][np.where(all_labels[indcs] == 0)[0]], names, np.arange(names.shape[0]), prefix = 'px ')
def look_at_features(names, indices):
N_INDCS = -1
map_test, map_train = map_overlay.basic_setup([100, 400], 50, label_name = "Jared")
clf = ObjectClassifier()
damage_labels = np.loadtxt('damagelabels50/Jared-3-3.csv', delimiter = ',').astype('int')
building_labels = np.loadtxt('damagelabels50/all_rooftops-3-3.csv', delimiter =',').astype('int')
segs = map_train.segmentations[50]
n_segs = int(np.max(segs.ravel()))+1
all_labels = np.zeros(n_segs) #Nothing = Green
all_labels[damage_labels] = 1 #Damage = Red
all_labels[building_labels] = 2 #Building = Blue
all_data, y = clf._get_X_y(map_train, 'Jared')
'''
clf.fit(map_train, label_name = 'Jared') #<-this one changesthe training for the importance order
names = clf.feat_names
print names
importances = clf.model.feature_importances_
indices = np.argsort(importances)[::-1]'''
pbar = custom_progress()
j = names.shape[0]
for i in pbar(indices[::-1]):
j-=1
fig = plt.figure(names[i])
plt.imshow(all_data[:,i][segs.astype('int')], cmap = 'gray')
fig.savefig('/Users/jsfrank/Desktop/features/{}-{}'.format(j, names[i]), format='png')
plt.close(fig)
def main_NZ():
model = ObjectClassifier()
map_2, map_3 = map_overlay.basic_setup()
labels = np.zeros(map_2.unique_segs(50).shape[0])
labels[np.loadtxt('damagelabels50/Jared-3-2.csv', delimiter = ',').astype('int')] = 1
probs = model.fit_and_predict(map_2, map_3, labels)
np.save('object_probs.npy', probs)
def _setup_map(self, random):
self.maps = [0,0,0,0]
self.tr, self.te = 2,3
self.maps[2:4] = list(map_overlay.basic_setup([100], 50))
if random:
self.building_rando = np.loadtxt('damagelabels50/non_damage_random-3-{}.csv'.format(self.tr)).astype('int')
else:
self.building_rando = np.loadtxt('damagelabels50/all_rooftops_random-3-{}.csv'.format(self.tr)).astype('int')
self.real_damage = np.loadtxt('damagelabels50/Jared-3-{}.csv'.format(self.tr), delimiter = ',').astype('int')
test_damage = np.loadtxt('damagelabels50/Jared-3-{}.csv'.format(self.te), delimiter = ',').astype('int')
test_segs = self.maps[self.te].segmentations[50].ravel().astype('int')
segs = self.maps[self.te].segmentations[50]
self.damage_ground = indcs2bools(test_damage, test_segs)[segs].ravel()
def transition():
N_INDCS = -1
map_test, map_train = map_overlay.basic_setup([100], 50, label_name = "Jared")
clf = ObjectClassifier()
damage_labels = np.loadtxt('damagelabels50/Jared-3-3.csv', delimiter = ',').astype('int')
building_labels = np.loadtxt('damagelabels50/all_buildings-3-3.csv', delimiter =',').astype('int')
segs = map_train.segmentations[50].ravel()
n_segs = int(np.max(segs))+1
all_labels = np.zeros(n_segs) #Nothing = Green
all_labels[damage_labels] = 1 #Damage = Red
all_labels[building_labels] = 2 #Building = Blue
all_data, y = clf._get_X_y(map_train, 'Jared')
clf.fit(map_train, label_name = 'Jared')
names = clf.feat_names
for i in range(50):
new_damage = np.concatenate((all_data[damage_labels], all_data[building_labels[:i*10]]), axis = 0)
print new_damage.shape
all_histos(new_damage, all_data[building_labels], all_data[np.where(all_labels == 0)[0]], names, [17], prefix = "{} ".format(i*10))
plt.show()
importances = self.model.feature_importances_
std = np.std([tree.feature_importances_ for tree in self.model.estimators_],
axis=0)
indices = np.argsort(importances)[::-1]
v_print("Feature ranking:", self.verbose)
for f in range(len(self.feat_names)):
v_print("{}. feature {}: ({})".format(f + 1, self.feat_names[indices[f]], importances[indices[f]]), self.verbose)
plt.figure()
plt.title("Feature importances")
plt.bar(range(len(self.feat_names)), importances[indices],
color="r", yerr=std[indices], align="center")
plt.xticks(range(len(self.feat_names)), self.feat_names[indices])
plt.xlim([-1, len(self.feat_names)])
if __name__ == "__main__":
map_2, map_3 = map_overlay.basic_setup()
model = PxClassifier()
real_damage = np.loadtxt('damagelabels50/Jared-3-3.csv', delimiter = ',').astype('int')
labels = map_3.mask_segments_by_indx(real_damage, 50, False)
labels = labels.reshape(map_3.shape[:2])
plt.imshow(labels)
plt.show()
probs = model.fit_and_predict(map_3, map_2, labels.ravel())
plt.imshow(probs.reshape(map_2.shape[:2]))
plt.show()
def visualize_noise(my_map):
img_num = my_map.name[-1]
for i in range(5):
noise_level = 10*i
img_name = 'damagelabels50/Sim_{}-3-{}.csv'.format(noise_level, img_num)
#plt.figure(img_name)
next_label = np.loadtxt(img_name, delimiter = ',')
next_img = my_map.mask_segments_by_indx(next_label, 50, with_img = True)
show_img(next_img)
img_name = 'damagelabels50/Joe-3-{}.csv'.format(img_num)
plt.figure(img_name)
next_label = np.loadtxt(img_name, delimiter = ',')
next_img = my_map.mask_segments_by_indx(next_label, 50, with_img = True)
plt.imshow(next_img)
img_name = 'damagelabels50/Jared-3-{}.csv'.format(img_num)
plt.figure(img_name)
next_label = np.loadtxt(img_name, delimiter = ',')
next_img = my_map.mask_segments_by_indx(next_label, 50, with_img = True)
plt.imshow(next_img)
plt.show()
if __name__ == '__main__':
map_train, map_test = map_overlay.basic_setup([100], 50, label_name = "Jared")
#main(map_test)
#main(map_train)
#main2(map_test)
#calc_percents()
visualize_noise(map_test)