Spyros

Hello. Thank you for your interest in our work. Yes, when learning the novel classes we freeze the feature extractor. The classification weight (of the classification layer) of each novel...

Yes. I will clean the corresponding code and I will upload it. I will let you know when is ready.

Did you build Caffe with cuDNN library? I think without it, Caffe uses much more GPU memory when applying the convolutional layers. That is probably why you run out of...

@litingfeng No, you do not need to free GPU memory after the first network. What you can do is in the demo_LocNet_object_detection_pipeline.m script to change the lines 90 and 91...

@litingfeng Regarding the `script_test_object_detection_pipeline_PASCAL.m`, it uses a single model (either the recognition or the localization model) at a time and that is why you do not have any problem running...

Apologies for the late answer. Due to a typo the files that I uploaded have as suffix .json instead of .h5. If you just renamed it yourself it should work...

Hello. Thanks for you interest! I am planning to release the code in the next 2-3 weeks.

Hi, I uploaded the code for the Image-Net based experiments. Soon, I will also upload the code for the MiniImageNet based experiments.

Hello. Thanks for you interest! I am planning to release the code in the next 2-3 weeks.

Hi, I uploaded the code for the Image-Net based experiments. Soon, I will also upload the code for the MiniImageNet based experiments.