Advanced 2d indexing in Theano to extract multiple pixels from an image

Question

Background

I have an image that I want to sample at the number (P) of x, y coordinates.

In Numpy, I can use advanced indexing for this:

 n_points = n_image[ [n_pos[:,1],n_pos[:,0]] ]

Returns a vector of pixels selected from an image.

Question

How can I do this advanced indexing in Theano?

What i tried

I tried the corresponding code in anano:

t_points = t_image[ [t_pos[:,1],t_pos[:,0]] ]

it is compilation and execution without any warning messages, but leads to an output form tensor (2,8,100), so it looks like it does some basic indexing option that returns a lot of image lines, instead of highlighting pixels.

Full code

import numpy as np
import theano.tensor as T
from theano import function, shared
import theano

P = 8 # Number of points to sample
n_image = np.zeros((100,100), dtype=np.int16)  # 100*100 image
n_pos = np.zeros( (P,2) , dtype=np.int32) # Coordinates within the image

# NUMPY Method
n_points = n_image[ [n_pos[:,1],n_pos[:,0]] ]

# THEANO method
t_pos = T.imatrix('t_pos')
t_image = shared( n_image )
t_points = t_image[ [t_pos[:,1],t_pos[:,0]] ]
my_fun = function( [t_pos], t_points)
t_points = my_fun(n_pos)

print n_points.shape
print t_points.shape

This prints (8,) for Numpy and (2,8,100) for Theano.

My version of Theano is 0.7.0.dev-RELEASE

Answer 1

, -, Theano numpy ( , ).

t_points = t_image[ [t_pos[:,1],t_pos[:,0]] ]

t_points = t_image[ (t_pos[:,1],t_pos[:,0]) ]

. numpy, .