CODEWITHSUNDEEP
pythonpickle
user8863227
user8863227

Reputation: 47

how to load pickle file to numpy array

I saved Numpy array to pickle file.The shape was (850,32,27).How can I load this pickle file to Numpy array? I tried to look up answer ,however I could not fine any.

Upvotes: 3

Views: 22614

Answers (5)

Gabriele Amorosino
Gabriele Amorosino

Reputation: 140

To open a pickle file using Numpy, you just have to specify the flag "allow_pickle=True".

Here is an example that should match your issue:

Create example data:

import numpy as np
data = np.zeros([850,32,27])

Save example data as pickle file:

import pickle
file='data.pkl'
with open(file, 'wb') as file:
      pickle.dump(data, file)

Load data using Numpy:

data_loaded = np.load(file,allow_pickle=True)

Check dimensions:

data_loaded.shape

BTW, as someone suggested, you can directly use pickle to load pickle file (that seems like the intuitive way):

Load data using Pickle:

with open(file, 'rb') as file:
    data_loaded = pickle.load(file)

Upvotes: 0

Steven
Steven

Reputation: 129

You can load pickled objects with np.load(), setting the parameter allow_pickle to True.

np.load('PATH_TO_PICKLE_FILE', allow_pickle=True)

For more info check this.

Upvotes: 6

GM1
GM1

Reputation: 491

You should load you array into a typle using the commands above, such as np.load('file', allow_pickle=True) suggested by Steven, or call:

filein0 = open('tensor_name', "rb")
t0 = pickle.load(filein0)

t0 is a tuple {tensor name, numpy array}

Then, your required numpy array can be extracted using simple:

arr = t0[1]

Upvotes: 1

Josh Friedlander
Josh Friedlander

Reputation: 11657

The canonical way of doing this is:

import pickle

with open('my_array.pickle', 'rb') as file:
    arr = pickle.load(file)

Upvotes: 0

Stein
Stein

Reputation: 3277

Numpy has a 

np.save()

https://docs.scipy.org/doc/numpy/reference/generated/numpy.save.html

and a 

np.load()

https://docs.scipy.org/doc/numpy/reference/generated/numpy.load.html#numpy.load

routine specifically for this purpose. Using pickle is possible, but it seems like over kill.

In [1]: import numpy as np                                                                                                                                                                                      

In [2]: A = np.linspace(0,1,10)                                                                                                                                                                                 

In [3]: np.save("bla.npy",A)                                                                                                                                                                                    

In [4]: B = np.load("bla.npy")                                                                                                                                                                                  

In [5]: B == A                                                                                                                                                                                                  
Out[5]: 
array([ True,  True,  True,  True,  True,  True,  True,  True,  True,
        True])

Upvotes: 0

Related Questions