To do deep learning projects, the original picture will generally be converted to other formats, to facilitate the learning framework for direct reading, such as Caffe's Leveldb, Lmdb,matlab's. m format, TensorFlow tfrecords format, etc.
Python reads and writes tfrecords format, please refer to http://blog.csdn.net/guvcolie/article/details/77460494
Python read and write Matlab, the. m format is also very simple, wrote a small example, directly on the code
Import Scipy.io as Sio
import numpy as NP
# to save the data variable in mat file at data1 corresponding location in
data = Np.array ([1, 2, 3, 5], dtype= ' Float32 ')
sio.savemat (' Result.mat ', {' data1 ': Data})
# # read out the corresponding data in the mat location in the file
= Data1 (' Result.mat ')
var = result[' data1 ']
print (VAR)
# Read the data format information in the mat file
info = sio.whosmat (' Result.mat ')
Print (info)