search

QVPR / Patch-NetVLAD

Code for the CVPR2021 paper "Patch-NetVLAD: Multi-Scale Fusion of Locally-Global Descriptors for Place Recognition"
MIT License
525 stars 74 forks source link

About the hard disk memory #19

Closed heng94 closed 3 years ago

heng94 commented 3 years ago

Hi, thanks for sharing your work. I have a question. After I ran the _featureextract.py, the size of produced files is over 800G, is it normal? I even cannot extract query images' features due to the lack of disk memory.

StephenHausler commented 3 years ago

Hi @Hanoch-ZH, yes that could be normal. If you are trying to run a really large dataset, like Nordland, that's expected for the performance version of our work. Couple options, you could use one of the smaller PCA models (modify performance.ini to change the model used), or use speed.ini.

Tobias-Fischer commented 3 years ago

storage.ini uses 128-d descriptors, compared to the default 4096-d descriptors. So that's a lot less storage required.

heng94 commented 3 years ago

@Tobias-Fischer @StephenHausler Thank you. I will try it and your work is really great!