Do we really need scene-specific pose encoders?

Yoli Shavit, Ron Ferens

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations


Visual pose regression models estimate the camera pose from a query image with a single forward pass. Current models learn pose encoding from an image using deep convolutional networks which are trained per scene. The resulting encoding is typically passed to a multi-layer perceptron in order to regress the pose. In this work, we propose that scene-specific pose encoders are not required for pose regression and that encodings trained for visual similarity can be used instead. In order to test our hypothesis, we take a shallow architecture of several fully connected layers and train it with pre-computed encodings from a generic image retrieval model. We find that these encodings are not only sufficient to regress the camera pose, but that, when provided to a branching fully connected architecture, a trained model can achieve competitive results and even surpass current state-of-the-art pose regressors in some cases. Moreover, we show that for outdoor localization, the proposed architecture is the only pose regressor, to date, consistently localizing in under 2 meters and 5 degrees.

Original languageEnglish
Title of host publicationProceedings of ICPR 2020 - 25th International Conference on Pattern Recognition
PublisherInstitute of Electrical and Electronics Engineers Inc.
Number of pages7
ISBN (Electronic)9781728188089
StatePublished - 2020
Externally publishedYes
Event25th International Conference on Pattern Recognition, ICPR 2020 - Virtual, Milan, Italy
Duration: 10 Jan 202115 Jan 2021

Publication series

NameProceedings - International Conference on Pattern Recognition
ISSN (Print)1051-4651


Conference25th International Conference on Pattern Recognition, ICPR 2020
CityVirtual, Milan

Bibliographical note

Publisher Copyright:
© 2020 IEEE


Dive into the research topics of 'Do we really need scene-specific pose encoders?'. Together they form a unique fingerprint.

Cite this