search

morgan1189 / hb-on-cnn-features-for-scenes

Hierarchical Bayes classifier working in the space of features produced by convolutional neural network for the images of scenes
2 stars 1 forks source link

Getting error when running in validation regime #1

Open Mahbubul-Alam-ODU opened 7 years ago

Mahbubul-Alam-ODU commented 7 years ago

Hi, I found your hb on CNN features code to be very interesting and nicely written. I was able to train (validate = 0) the model using the mat file: "scenes_hbn_data.mat". However, when I try to run the code in validation regime (validate=1) using the mat file "scenes_hbn_data_validation.mat", I get the "b_cat" value always = 1. I debugged the code and found that when a validation input is provided in this function: "sample_category_most_likely_cat_id(hbn_tree, x)" the "posteriror" values for all "c" are NaNs. Am I doing some thing wrong? Any feedback is highly appreciated.

vikramreddy0307 commented 4 years ago

Hey, Can you please tell me how did you train it? All I see is functions declaration, how should I execute them? Regards Vikram