search

cvjena / part_constellation_models

Code for the ICCV 2015 paper "Neural Activation Constellations: Unsupervised Part Model Discovery with Convolutional Networks."
47 stars 22 forks source link

Error using caffe_prepare_image (line 25) Subscript indices must either be real positive integers or logicals. #3

Open pcjeff opened 7 years ago

pcjeff commented 7 years ago

Hi, I am trying to run training process by the following command. start('basedir','/path/to/CUB_200_2011_dataset/','cnn_dir','./cnn_finetuning/vgg19/','imagedir','/path/to/CUB_200_2011_dataset/images');.

However, I got the error below:


Starting parallel pool (parpool) using the 'local' profile ... "/auto/matlab-2015b/toolbox/distcomp/bin/util/psname.sh" GLNXA64 1074: Aborted
Image 11788: 200.Common_Yellowthroat/Common_Yellowthroat_0055_190967.jpg
Error using caffe_prepare_image (line 25)
Subscript indices must either be real positive integers or logicals.

Error in caffe_gradients (line 39)
    im = caffe_prepare_image(im,mean,width);

Error in parts_locs_from_grads (line 25)
    parfor (i = 1:size(imagelist,1), opts.parfor_arg)

Error in start (line 162)
    part_locs = parts_locs_from_grads(opts);

I generated the txt files by following the tutorial. Have you ever run into this problem?

MarcelSimon commented 7 years ago

I didn't have this problem. You can try replacing the parfor loops by regular for loops for easier debugging.