search

ovysotska / image_sequence_localizer

The framework for visual place recognition in outdoor environments using image sequences.
46 stars 11 forks source link

Localization with images as input #14

Closed ovysotska closed 1 year ago

ovysotska commented 1 year ago

This PR adds a netvlad network that reads the pre-trained weights. The network is used to extract features from images.

And now, YOU ONLY NEED IMAGES to run the code :partying_face:

P.S. Also this PR restructures the pipeline to be able to to use the code as before.