ekingedik/loupe-pytorch

Learnable mOdUle for Pooling fEatures(LOUPE) implementation in pytorch and it's example use

loupe-pytorch

Learnable mOdUle for Pooling fEatures(LOUPE) implementation in pytorch and it's example use Converted the code from Antoine Miech, Ivan Laptev, Josef Sivic presented in https://github.com/antoine77340/LOUPE Inspired from https://gist.github.com/catta202000/cbf5cae9199ccd72adddef452358df3b

The code is not extensively tested, please ping me if you notice any problems.

LSTM-Loupe shows the use of NetVLAD layer for pooling the outputs of multiple LSTM's.

loupe-pytorch includes the pytorch implementations of multiple learnable pooling layers: NetVLAD, NetRVLAD, SoftDBoW and NetFV NetFV implementation is currently not working.

loupe-pytorch's Issues

NetVLAD

when I use NetVLAD, I find that this model get reshaped_input in this way:

reshaped_input = input.view(-1, self.feature_size)

but image is channel first in pytorch, like the screenshot below, although the above operation can get the same shape tensor, this tensor is wrong