File size: 1,388 Bytes
bbde80b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
import os
import sys
import pickle
import numpy as np
import random
sys.path.insert(0, os.getcwd())
from lib.utils.tools import read_pkl
from lib.data.datareader_h36m import DataReaderH36M
from tqdm import tqdm
def save_clips(subset_name, root_path, train_data, train_labels):
len_train = len(train_data)
save_path = os.path.join(root_path, subset_name)
if not os.path.exists(save_path):
os.makedirs(save_path)
for i in tqdm(range(len_train)):
data_input, data_label = train_data[i], train_labels[i]
data_dict = {
"data_input": data_input,
"data_label": data_label
}
with open(os.path.join(save_path, "%08d.pkl" % i), "wb") as myprofile:
pickle.dump(data_dict, myprofile)
datareader = DataReaderH36M(n_frames=243, sample_stride=1, data_stride_train=81, data_stride_test=243, dt_file = 'h36m_sh_conf_cam_source_final.pkl', dt_root='data/motion3d/')
train_data, test_data, train_labels, test_labels = datareader.get_sliced_data()
print(train_data.shape, test_data.shape)
assert len(train_data) == len(train_labels)
assert len(test_data) == len(test_labels)
root_path = "data/motion3d/MB3D_f243s81/H36M-SH"
if not os.path.exists(root_path):
os.makedirs(root_path)
save_clips("train", root_path, train_data, train_labels)
save_clips("test", root_path, test_data, test_labels)
|