The dataset and code presented here is introduced in the paper "Deep Learning Features at Scale for Visual Place Recognition" that has been accepted by ICRA 2017: https://arxiv.org/abs/1701.05105.
Should you have any question, please write an email to firstname.lastname@example.org
The SPED dataset will go on-line soon.
1) The HybridNet in the paper:
2) AmosNet in the paper: