Hi.
Thanks so much for publishing this awesome tool. I have two quick questions about model architectures:
-
It looks like the model on this repo (in Models/DBN_model.h5) is based on a VGG16 architecture. This also seems to be true for the "DeepBrainNet_Normalized_General" models you share on your box link (in the OTHER folder). These VGG models are working well for me, but I know that in the paper you suggest that the inceptionresnet -v2 model seems to outperform other architectures. Would you (apriori) recommend one architecture over the others?
-
If you do recommend the inceptionresnet-v2 achitecture, is there a way for me to access the h5 file associated with that model? I see the DeepBrainNet_InceptionResnetv2.h5 file inside the OTHER folder on box; however, I have been unable to successfully load this h5 file. I am able to load the other h5 files in this folder, so I wonder if that file is perhaps corrupted? Or is there some reason I wouldn't be able to load it with a typical call to keras.models.load_model()?
Thanks,
Amy
Hi.
Thanks so much for publishing this awesome tool. I have two quick questions about model architectures:
It looks like the model on this repo (in Models/DBN_model.h5) is based on a VGG16 architecture. This also seems to be true for the "DeepBrainNet_Normalized_General" models you share on your box link (in the OTHER folder). These VGG models are working well for me, but I know that in the paper you suggest that the inceptionresnet -v2 model seems to outperform other architectures. Would you (apriori) recommend one architecture over the others?
If you do recommend the inceptionresnet-v2 achitecture, is there a way for me to access the h5 file associated with that model? I see the DeepBrainNet_InceptionResnetv2.h5 file inside the OTHER folder on box; however, I have been unable to successfully load this h5 file. I am able to load the other h5 files in this folder, so I wonder if that file is perhaps corrupted? Or is there some reason I wouldn't be able to load it with a typical call to keras.models.load_model()?
Thanks,
Amy