search

QVPR / Patch-NetVLAD

Code for the CVPR2021 paper "Patch-NetVLAD: Multi-Scale Fusion of Locally-Global Descriptors for Place Recognition"
MIT License
510 stars 72 forks source link

Reproduce the results of DELG in the paper #50

Closed hellocasper closed 2 years ago

hellocasper commented 2 years ago

Hi, thank you for sharing this great work. In the official paper, DELG achieves the top performance on some datasets. May I know how to reproduce those results of DELG? Do you use the pre-trained model or fine-tune it on Pitts30k/MSLS?

Tobias-Fischer commented 2 years ago

Hi @hellocasper, we use the pre-trained model from the authors' repository: https://github.com/tensorflow/models/blob/master/research/delf/delf/python/delg/DELG_INSTRUCTIONS.md

Hope that helps.

hellocasper commented 2 years ago

Hi @Tobias-Fischer, thanks for your help and quick reply! Should I transfer the model to PyTorch first, as I notice the official repository is based on TensorFlow?

Tobias-Fischer commented 2 years ago

We used the official repository without any changes (i.e. we used tensorflow) for the DELG comparisons.

hellocasper commented 2 years ago

OK, thanks! @Tobias-Fischer