Animal identification in low quality camera-trap images using very deep convolutional neural networks and confidence thresholds

Alexander Gomez-Villa, German Diez, Augusto Salazar*, Angelica Diaz

*Autor corresponent d’aquest treball

Producció científica: Capítol de llibreCapítolRecercaAvaluat per experts

45 Cites (Scopus)

Resum

Monitoring animals in the wild without disturbing them is possible using camera trapping framework. Automatic triggered cameras, which take a burst of images of animals in their habitat, produce great volumes of data, but often result in low image quality. This high volume data must be classified by a human expert. In this work a two step classification is proposed to get closer to an automatic and trustfully camera-trap classification system in low quality images. Very deep convolutional neural networks were used to distinguish images, firstly between birds and mammals, secondly between mammals sets. The method reached 97.5%97.5% and 90.35%90.35% in each task. An alleviation mode using a confidence threshold of automatic classification is proposed, allowing the system to reach 100%100% of performance traded with human work.
Idioma originalAnglès
Títol de la publicacióAdvances in Visual Computing - 12th International Symposium, ISVC 2016, Proceedings
EditorsGeorge Bebis, Bahram Parvin, Sandra Skaff, Daisuke Iwai, Richard Boyle, Darko Koracin, Fatih Porikli, Carlos Scheidegger, Alireza Entezari, Jianyuan Min, Amela Sadagic, Tobias Isenberg
Pàgines747-756
Nombre de pàgines10
DOIs
Estat de la publicacióPublicada - 10 de des. 2016
Publicat externament

Sèrie de publicacions

NomLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volum10072 LNCS
ISSN (imprès)0302-9743
ISSN (electrònic)1611-3349

Fingerprint

Navegar pels temes de recerca de 'Animal identification in low quality camera-trap images using very deep convolutional neural networks and confidence thresholds'. Junts formen un fingerprint únic.

Com citar-ho