chaojianyu/hierarchical-bilinear-pooling

Implementation for <Hierarchical Bilinear Pooling for Fine-Grained Visual Recognition> in ECCV'18.

hierarchical-bilinear-pooling's Issues

do you train on resent models?

Hi @ChaojianYu

Bilinear pooling works well on VGG16, but do you ever train on resnet model? like resnet50.

Thanks.

about the sum pooling layer

hi @ChaojianYu
what is the difference between the normal pooling layer and your sum pooling layer?
thanks

缺少log_layer.cpp文件

About loss of training with ft_all.prototxt

I train the HBP net with ft_all_prototxt and the premodel, but the loss cannot come down?

input image size

About Visualization?

Could you please tell me how to implement the visualization operation in section 4.4 [Qualitative Visualization] of your paper?
Please tell me. I'm new in this area. Thank you very much.

About test phase image

First, thank you about the HBP model, I have some question about how do you crop image about test image in CUB. You said that 'Only center cropping is involved during inference', but it seems that you should first resize your test image to some size right? What size is it? is it 512512 and then center crop a 448448?

Train_accuracy problems

hi,@ChaojianYu, Recently, i found a problem.when i delete the include: phase: TEST in last accuracy layer of ft_last_layer.prototxt ,the loss will be in oscillation？Maybe I made some mistakes. Can you tell me why? i am so confused. Look forward to you reply.

About the result

@ChaojianYu sorry to trouble but I can't get the accuracy you have mentioned in your paper just through this project on the CUB dataset.
So I wonder if you use some tricks during your training, could you please tell me something interesting?
Thanks a lot.

谁有pytorch的实现吗，达到作者的精度的

Please use the caffe-users list for usage, installation, or modeling questions, or other requests for help.
Do not post such requests to Issues. Doing so interferes with the development of Caffe.

Please read the guidelines for contributing before submitting this issue.

Issue summary

Steps to reproduce

If you are having difficulty building Caffe or training a model, please ask the caffe-users mailing list. If you are reporting a build error that seems to be due to a bug in Caffe, please attach your build configuration (either Makefile.config or CMakeCache.txt) and the output of the make (or cmake) command.

Your system configuration

Operating system:
Compiler:
CUDA version (if applicable):
CUDNN version (if applicable):
BLAS:
Python or MATLAB version (for pycaffe and matcaffe respectively):