Comparison of human eye and artificial intelligence in identifying Eurasian lynx individuals from camera trap photos

master's thesis

• Title (croatian): Usporedba ljudskog oka i umjetne inteligencije pri individualnom raspoznavanju euroazijskog risa na fotografijama s fotozamki
• Author: Jana Ayman Fathi Emara
• Mentor: Magda Sindičić (mentor)
• Mentor: Tomislav Gomerčić (mentor)
• Committee member: Alen Slavica (predsjednik povjerenstva)
• Committee member: Ira Topličanec (član povjerenstva)
• Committee member: Magda Sindičić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Veterinary Medicine; Zagreb
• Defense date and country: 2024-11-19, Croatia
• Scientific / art field, discipline and subdiscipline: BIOMEDICINE AND HEALTHCARE; Veterinary Medicine; Animal Production and Biotechnology

Abstract

The monitoring of the Eurasian lynx (Lynx lynx) population is not an easy task due to
the shy nature of this predator species as well as the vast territory they inhabit. In an attempt to
enrich this field of wildlife research, artificial intelligence algorithms have been developed by
experts to aid in the identification process. However, these algorithms such as MiewID deep
learning technology used by the Whiskerbook program must be checked for dependability. A
total of 112 images from the Faculty of Veterinary Medicine University in Zagreb’s database
were uploaded to Whiskerbook for analysis by an expert observer as well as by an amateur for
individual identification. These images were captured by camera traps strategically placed in
areas of lynx distribution in Croatia as part of the LIFE lynx project. The expert using visual
identification, the expert using MiewID and the amateur using MiewID were able to identify
104, 85 and 75 images out of the 112, respectfully. These 112 images were found to be
comprised of 29 different individual animals. The expert using visual identification, expert
using MiewID, and the amateur using MiewID succeeded in identifying 28, 23, and 20
individual animals respectfully. The main advantage that the artificial intelligence provided
was the detection of man-made errors. However, the experienced human eye is still the more
accurate and precise method of Eurasian lynx identification.

Abstract (croatian)

Praćenje populacija Eurozaijskog risa (Lynx lynx) je zahtjevno zbog skrovitog načina života i
rasprostranjenosti na velikim površinama. Stoga se danas sve više koristi umjetna inteligencija
kako bi se olakšalo istraživanje divljih životinja i individualna identifikacija jedinki. Cilj ovog
diplomskog rada bio je istražiti pouzdanost algoritma MiewID kojeg koristi program
Whiskerbook za individualnu identifikaciju risa. Usporedili smo uspješnost identifikacije
jedinki risa na 112 fotografija prikupljenih pomoću fotozamki postavljenih na području
rasprostranjenosti risa u Hrvatskoj, u sklopu LIFE Lynx projekta. Stručnjak je pomoću vizualne
identifikacije uspješno identificirao 104 fotografije, dok je stručnjak uz pomoću algoritma
MiewID uspješno analizirao 85 fotografija, a amater pomoću algortitma MiewID je analizirao
75 fotografija od istraženih 112. Na fotografijama su zabilježene 29 jedinke risa, od koji je
stručnjak prepoznao 28, stručnjak uz pomoć algoritma 23, a amater uz pomoć algoritma 20
jedinki. Najveća prednost upotrebe umjetne inteligencije je u otkrivanju pogreški koje je
napravio prilikom vizualne identifikacije, no vizualna identifikacija stručnjaka je je bila
uspješnija od algoritma.

• Keywords: camera traps; MiewID; Whiskerbook; artificial intelligence; identification; Croatia; eurasian lynx; Lynx lynx
• Keywords (croatian): fotozamke; MiewID; Whiskerbook; umjetna inteligencija; identifikacija; Hrvatska; euroazjski ris; Lynx lynx
• Language: english
• URN:NBN: urn:nbn:hr:178:510073
• Study programme: Title: Veterinary Medicine; Study programme type: university; Study level: integrated undergraduate and graduate; Academic / professional title: doktor/doktorica veterinarske medicine (doktor/doktorica veterinarske medicine)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Access restricted to students and staff of home institution
• Created on: 2025-01-07 14:15:49