Blur mapping using semantic segmentation

Hello Everyone, I am working on a project that includes blur mapping and blur scoring. My task is to generate a map which is black colored only for the area that is blurred in the original image. I am using simple semantic segmentation process with 2 classes (blur and not blur). I am using Vggnet for semantic segmentation in MATLAB 2017b. This is the Code,

This is the error I am getting, Error using nnet.internal.cnn.util.TrainNetworkDataValidator/assertCorrectResponseSizeForOutputLayer (line 217) Invalid training data. The output size (2) of the last layer doesn't match the number of classes (2). Please help me debug this error!! Thanks.

Answer 1

[kobe liang]

i had this problem,too,which i think link to conv1x1,but i still not found any solution.