Exploring loss functions for optimising the accuracy of Siamese Neural Networks in Re-Identification applications

Jonathan Eichild Schmidt, Oscar Edvard Mäkinen, Simon Gørtz Flou Nielsen, Anders Skaarup Johansen, Kamal Nasrollahi, Thomas B. Moeslund

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

Abstrakt

As Re-Identification (Re-ID) is becoming more and more common in today’s world, the need for more optimizedalgorithms also becomes more wanted. This is due to the importance of high accuracy as the consequences ofan incorrect match can mean security issues, if used to gain access or result in incorrect findings in science dueto wrong data. This paper explores enhancing the performance of Siamese Neural Networks by exploring theperformance of loss functions to better suit the user’s Re-IDing needs. These loss functions are Triplet loss,Triplet Hard loss and Quadruplet loss. Results show that the Triplet hard loss function performs better thanthe two others. The functions were tested on a human dataset as well as on animal datasets
OriginalsprogEngelsk
TitelInternational Conference on Machine Vision
Antal sider7
UdgivelsesstedSPIE Digital Library
ForlagSPIE - International Society for Optical Engineering
StatusAccepteret/In press - 2021

Fingeraftryk

Dyk ned i forskningsemnerne om 'Exploring loss functions for optimising the accuracy of Siamese Neural Networks in Re-Identification applications'. Sammen danner de et unikt fingeraftryk.

Citationsformater