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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Visual Place Recognition (VPR) aims to retrieve frames from a geotaggeddatabase that are located at the same place as the query frame. To improve therobustness of VPR in perceptually aliasing scenarios, sequence-based VPRmethods are proposed. These methods are either based on matching between framesequences or extracting sequence descriptors for direct retrieval. However, theformer is usually based on the assumption of constant velocity, which isdifficult to hold in practice, and is computationally expensive and subject tosequence length. Although the latter overcomes these problems, existingsequence descriptors are constructed by aggregating features of multiple framesonly, without interaction on temporal information, and thus cannot obtaindescriptors with spatio-temporal discrimination. In this paper, we propose asequence descriptor that effectively incorporates spatio-temporal information.Specifically, spatial attention within the same frame is utilized to learnspatial feature patterns, while attention in corresponding local regions ofdifferent frames is utilized to learn the persistence or change of featuresover time. We use a sliding window to control the temporal range of attentionand use relative position encoding to construct sequential relationshipsbetween different features. This allows our descriptors to capture theintrinsic dynamics in a sequence of frames. Comprehensive experiments onchallenging benchmark datasets show that the proposed approach outperformsrecent state-of-the-art methods.

Learning Sequence Descriptor based on Spatio-Temporal Attention for Visual Place Recognition