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

This paper combines deep learning techniques for species detection, 3D modelfitting, and metric learning in one pipeline to perform individual animalidentification from photographs by exploiting unique coat patterns. This is thefirst work to attempt this and, compared to traditional 2D bounding box orsegmentation based CNN identification pipelines, the approach provideseffective and explicit view-point normalisation and allows for a straightforward visualisation of the learned biometric population space. Note that dueto the use of metric learning the pipeline is also readily applicable to openset and zero shot re-identification scenarios. We apply the proposed approachto individual Grevy's zebra (Equus grevyi) identification and show in a smallstudy on the SMALST dataset that the use of 3D model fitting can indeed benefitperformance. In particular, back-projected textures from 3D fitted modelsimprove identification accuracy from 48.0% to 56.8% compared to 2D bounding boxapproaches for the dataset. Whilst the study is far too small accurately toestimate the full performance potential achievable in larger-scale real-worldapplication settings and in comparisons against polished tools, our work laysthe conceptual and practical foundations for a next step in animal biometricstowards deep metric learning driven, fully 3D-aware animal identification inopen population settings. We publish network weights and relevant facilitatingsource code with this paper for full reproducibility and as inspiration forfurther research.

Towards Individual Grevy's Zebra Identification via Deep 3D Fitting and Metric Learning