scipy.sparse.issparse(v) object has no attribute 'sparse'

[willard-yuan]

Hi, pixelogik. I'm very interested in the Local Sensitive Hashing and plan to use it for image retrieval. I install the NearPy and the redis packages. Then I test the following example:

import numpy
from nearpy import Engine
from nearpy.hashes import RandomBinaryProjections

# Dimension of our vector space
dimension = 500

# Create a random binary hash with 10 bits
rbp = RandomBinaryProjections('rbp', 10)

# Create engine with pipeline configuration
engine = Engine(dimension, lshashes=[rbp])

# Index 1000000 random vectors (set their data to a unique string)
for index in range(100000):
    v = numpy.random.randn(dimension)
    engine.store_vector(v, 'data_%d' % index)

# Create random query vector
query = numpy.random.randn(dimension)

# Get nearest neighbours
N = engine.neighbours(query)

It throws out the error:

C:\Python27\python.exe D:/python/projects/lsh/lsh-image-search.py
Traceback (most recent call last):
  File "D:/python/projects/lsh/lsh-image-search.py", line 17, in <module>
    engine.store_vector(v, 'data_%d' % index)
  File "C:\Python27\lib\site-packages\nearpy\engine.py", line 79, in store_vector
    for bucket_key in lshash.hash_vector(v):
  File "C:\Python27\lib\site-packages\nearpy\hashes\randombinaryprojections.py", line 65, in hash_vector
    if scipy.sparse.issparse(v):
AttributeError: 'module' object has no attribute 'sparse'

This happens because the scipy module doesn't have any attribute named sparse. That attribute only gets defined when you import scipy.sparse.

So in the randombinaryprojections.py, it's good to import scipy.sparse. After I do this, it succeed.

[pixelogik]

Hi there :)

Thanks for the hint, I added the imports to all hashes with the last commit.

Funny, I started with the library because of image retrieval, but never did the actual application. Good luck and fun with that!

Cheers