English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

An important challenge for autonomous agents such as robots is to maintain aspatially and temporally consistent model of the world. It must be maintainedthrough occlusions, previously-unseen views, and long time horizons (e.g., loopclosure and re-identification). It is still an open question how to train sucha versatile neural representation without supervision. We start from the ideathat the training objective can be framed as a patch retrieval problem: givenan image patch in one view of a scene, we would like to retrieve (with highprecision and recall) all patches in other views that map to the samereal-world location. One drawback is that this objective does not promotereusability of features: by being unique to a scene (achieving perfectprecision/recall), a representation will not be useful in the context of otherscenes. We find that it is possible to balance retrieval and reusability byconstructing the retrieval set carefully, leaving out patches that map tofar-away locations. Similarly, we can easily regulate the scale of the learnedfeatures (e.g., points, objects, or rooms) by adjusting the spatial tolerancefor considering a retrieval to be positive. We optimize for (smooth) AveragePrecision (AP), in a single unified ranking-based objective. This objectivealso doubles as a criterion for choosing landmarks or keypoints, as patcheswith high AP. We show results creating sparse, multi-scale, semantic spatialmaps composed of highly identifiable landmarks, with applications in landmarkretrieval, localization, semantic segmentation and instance segmentation.

LoCUS: Learning Multiscale 3D-consistent Features from Posed Images