Image Segmentation on Pascal VOC - Structure of X in data_train

121 views
Skip to first unread message

Abhinandan Dubey

unread,
Mar 13, 2017, 1:58:47 AM3/13/17
to pystruct
Hi,

I am going to use CRF for yeast cells image segmentation. I was trying to understand what should be fed into the library functions for them to work.

I tried seeing around the shapes and inferring what goes inside - but tough luck.


print(np.array(data_train['X']).shape)
print(np.array(data_train['X'])[0][0].shape)
print(np.array(data_train['X'])[0][1].shape)
print(np.array(data_train['X'])[0][2].shape)

Output:

(964, 3)
(92, 21)
(238, 2)
(238, 3)

I understand that Dr. Mueller is very busy and hasn't been able to document the code, can someone just tell me what is inside the variable X in the dict - data_train !

Thanks for your time !

Abhinandan Dubey

unread,
Mar 13, 2017, 4:47:44 PM3/13/17
to pystruct

JL Meunier

unread,
Mar 14, 2017, 6:20:32 AM3/14/17
to pystruct
Hi

I do not know this dataset, but most probably you have:
964 graphs. Each graph is a triplet.
The first one has 92 nodes, each node is described by 21 features. This graph has 238 edges (the 238x2 array), each described by 3 features. 


JL
Reply all
Reply to author
Forward
0 new messages