def save_training(self):
'''save object in hdf5 file format
Args:
filename: str
path to the hdf5 file containing the saved object
'''
if '.hdf5' in self.filename:
save_dict_to_hdf5(self.__dict__, self.filename)
else:
raise Exception("Filename not supported")
def create_pos_dataset(pos, target_size, out_dir):
ims = []
landmarks = []
labels = []
for line in pos:
words = line.strip().split()
if '.jpg' in words[0]:
image_file_name = dealdata + '/' + words[0]
else:
image_file_name = dealdata + '/' + words[0] + '.jpg'
im = cv2.imread(image_file_name)
im = resize(im, target_size)
im = im.astype('uint8')
ims.append(im)
labels.append(int(words[1]))
landmark = words[2:6]
landmark = list(map(float, landmark))
landmark = np.array(landmark, dtype='float32')
landmarks.append(landmark)
if len(ims) % 500 == 0:
print('pos data doing, total: {}'.format(len(ims)))
landmark_data = list(zip(labels, ims, landmarks))
random.shuffle(landmark_data)
labels, ims, landmarks = zip(*landmark_data)
save_dict_to_hdf5({
'labels': labels,
'ims': ims,
'bbox': landmarks
}, out_dir)
print('pos data done, total: {}'.format(len(ims)))
def create_landmark_dataset(net_name, landmark_anno, target_size, out_dir):
ims = []
landmarks = []
labels = []
for line in landmark_anno:
words = line.strip().split()
if '.jpg' in words[0]:
image_file_name = words[0]
else:
image_file_name = words[0] + '.jpg'
im = cv2.imread(image_file_name)
im = resize(im, target_size)
im = im.astype('uint8')
ims.append(im)
labels.append(int(words[1]))
landmark = words[2:12]
landmark = list(map(float, landmark))
landmark = np.array(landmark, dtype='float32')
landmarks.append(landmark)
if len(ims) % 500 == 0:
print('landmarks data doing, total: {}'.format(len(ims)))
landmark_data = list(zip(labels, ims, landmarks))
random.shuffle(landmark_data)
labels, ims, landmarks = zip(*landmark_data)
landmark_data_filename = os.path.join(out_dir, 'landmarks_shuffle.h5')
save_dict_to_hdf5({
'labels': labels,
'ims': ims,
'landmarks': landmarks
}, landmark_data_filename)
print('landmarks data done, total: {}'.format(len(ims)))
frame = frame/float(255) # normalize
frame = frame.astype('float32')
# cv2.imshow('frame',frame)
if gray: frame = np.tile(frame,(1,1,1)) # give it the channel dim
video.append(frame)
# import time
# time.sleep(0.5)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
video = u.stack(video)
action_vids[vid] = video # video, subsampled, evenly spaced, consecutive video
pbar.finish()
u.save_dict_to_hdf5(dataset=action_vids, dataset_name=action+'_subsamp='+str(subsample), dataset_folder=out)
# action_vids = np.vstack(action_vids) # consecutive video ACTUALLY THIS MIGHT NOT BE TRUE. WE NEED THE VIDEOS TO BE SEPARATE!
# randomly permute -- don't do this!
# tm1s = np.random.permutation(range(0,len(action_vids)-1,2))
# ts = np.array([i+1 for i in tm1s])
# shuffle_idxs = list(it.next() for it in itertools.cycle([iter(tm1s), iter(ts)])) # groups of 2
# action_vids = action_vids[np.array(shuffle_idxs),:,:]
# action_data[action] = action_vids
# save
# u.save_dict_to_hdf5(action_data, 'actions_2_frame_subsample_' + str(subsample), root)
(1, 1, 1)) # give it the channel dim
video.append(frame)
# import time
# time.sleep(0.5)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
video = u.stack(video)
action_vids[
vid] = video # video, subsampled, evenly spaced, consecutive video
pbar.finish()
u.save_dict_to_hdf5(dataset=action_vids,
dataset_name=action + '_subsamp=' + str(subsample),
dataset_folder=out)
# action_vids = np.vstack(action_vids) # consecutive video ACTUALLY THIS MIGHT NOT BE TRUE. WE NEED THE VIDEOS TO BE SEPARATE!
# randomly permute -- don't do this!
# tm1s = np.random.permutation(range(0,len(action_vids)-1,2))
# ts = np.array([i+1 for i in tm1s])
# shuffle_idxs = list(it.next() for it in itertools.cycle([iter(tm1s), iter(ts)])) # groups of 2
# action_vids = action_vids[np.array(shuffle_idxs),:,:]
# action_data[action] = action_vids
# save
# u.save_dict_to_hdf5(action_data, 'actions_2_frame_subsample_' + str(subsample), root)