file_shape_check.py

#이미지의 shape 찍기
import PIL
import torchvision.transforms as transforms
import numpy as np
import matplotlib.pyplot as plt

'''img = PIL.Image.open("/home/kwon/sparse/data/custom/image/filename_time.struct_time(tm_year=2022, tm_mon=7, tm_mday=18, tm_hour=5, tm_min=18, tm_sec=12, tm_wday=0, tm_yday=199, tm_isdst=0).jpg")
numpyimg = np.array(img, dtype='f4')
print(numpyimg)
print(numpyimg.max())
print(numpyimg.shape)

tf = transforms.ToTensor()
img_t = tf(img)

print(img_t)
print(img_t.max())
print(img_t.size())'''

#npy의 shape 찍기
y = np.load('/home/kwon/sparse/data/custom/train/filename_time.struct_time(tm_year=2022, tm_mon=7, tm_mday=19, tm_hour=2, tm_min=27, tm_sec=23, tm_wday=1, tm_yday=200, tm_isdst=0).png.npy')
npy = np.array(y, dtype='f4')
print(npy)
print(npy.shape)

#h5 파일의 shape 찍기
'''import h5py
import torch
import numpy as np

f = h5py.File('/home/kwon/data/nyudepthv2/train/basement_0001a/00001.h5', 'r')
file = np.array(f['depth'])
print(file)

#pil_file = PIL.Image.fromarray(file)
#pil_file.show()

plt.imshow(file)
plt.show()

torch_file = torch.from_numpy(file)
print(torch_file)
print(torch_file.max())
print(torch_file.size())'''

#file의 NaN값 존재여부 확인
'''y = np.load('/home/kwon/sparse/data/custom/depth/filename_time.struct_time(tm_year=2022, tm_mon=7, tm_mday=28, tm_hour=1, tm_min=39, tm_sec=49, tm_wday=3, tm_yday=209, tm_isdst=0).png.npy', allow_pickle=True)
npy = np.array(y, dtype='f4')
#plt.imshow(npy)
print(np.isnan(npy).sum())
print(np.isinf(npy).sum())
print(npy)
print(npy.shape)'''