Unsupervised Place Recognition with Deep Embedding Learning over Radar Videos
This addresses the problem of robust place recognition in challenging conditions like radar data for applications such as autonomous navigation, offering a novel unsupervised approach that outperforms supervised methods.
The paper tackles place recognition using sequences of radar images by learning an embedding in an unsupervised manner, achieving 98.38% correct localization with just the nearest database candidate on 280 km of data.
We learn, in an unsupervised way, an embedding from sequences of radar images that is suitable for solving place recognition problem using complex radar data. We experiment on 280 km of data and show performance exceeding state-of-the-art supervised approaches, localising correctly 98.38% of the time when using just the nearest database candidate.