search

hbilen / WSDDN

Weakly Supervised Deep Detection Networks (CVPR 2016)
148 stars 53 forks source link

When run cnn_wsdnn_train.m meet the problem #9

Closed yanxp closed 6 years ago

yanxp commented 6 years ago

Error using cellfun Input #2 expected to be a cell array, was struct instead.

Error in prepare_wsddn (line 13) relu6p = find(cellfun(@(a) strcmp(a.name, 'relu6'), net.layers)==1);

Error in cnn_wsddn_train (line 60) net = prepare_wsddn(net,nopts);

drcege commented 6 years ago

@yanxp I know the reason. In the training phase, you should use imagenet-vgg-f.mat instead of vggf-edgeboxes-boxsc-spatreg.mat. After loading imagenet-vgg-f.mat, its layers field is a 1x21 cell array.

alexshaodong commented 5 years ago

Error using cellfun Input #2 expected to be a cell array, was struct instead.

Error in prepare_wsddn (line 13) relu6p = find(cellfun(@(a) strcmp(a.name, 'relu6'), net.layers)==1);

Error in cnn_wsddn_train (line 60) net = prepare_wsddn(net,nopts);

Hello! If I want to train with my own data, how to generate edgeboxes mat files, similar to EdgeBoxes VOC2007trainval. mat and EdgeBoxes VOC2007test. mat files.

alexshaodong commented 5 years ago

@yanxp I know the reason. In the training phase, you should use imagenet-vgg-f.mat instead of vggf-edgeboxes-boxsc-spatreg.mat. After loading imagenet-vgg-f.mat, its layers field is a 1x21 cell array.

Hello! If I want to train with my own data, how to generate edgeboxes mat files, similar to EdgeBoxes VOC2007trainval. mat and EdgeBoxes VOC2007test. mat files.