prikaz prve stranice dokumenta A comparison of accuracy of image recognition apps for identification of edible weed species
  Download
PDF 1.39 MB
master's thesis
A comparison of accuracy of image recognition apps for identification of edible weed species
Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek, 2023. urn:nbn:hr:151:866118
Grgić, Domagoj
Josip Juraj Strossmayer University of Osijek
Faculty of Agrobiotechnical Sciences Osijek
Department of Phytomedicine

Institutional repository: Repository of the Faculty of Agrobiotechnical Sciences Osijek

Cite this document

Grgić, D. (2023). A comparison of accuracy of image recognition apps for identification of edible weed species (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek. Retrieved from https://urn.nsk.hr/urn:nbn:hr:151:866118

Grgić, Domagoj. "A comparison of accuracy of image recognition apps for identification of edible weed species." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek, 2023. https://urn.nsk.hr/urn:nbn:hr:151:866118

Grgić, Domagoj. "A comparison of accuracy of image recognition apps for identification of edible weed species." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek, 2023. https://urn.nsk.hr/urn:nbn:hr:151:866118

Grgić, D. (2023). 'A comparison of accuracy of image recognition apps for identification of edible weed species', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek, accessed 10 July 2024, https://urn.nsk.hr/urn:nbn:hr:151:866118

Grgić D. A comparison of accuracy of image recognition apps for identification of edible weed species [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek; 2023 [cited 2024 July 10] Available at: https://urn.nsk.hr/urn:nbn:hr:151:866118

D. Grgić, "A comparison of accuracy of image recognition apps for identification of edible weed species", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Agrobiotechnical Sciences Osijek, Osijek, 2023. Available at: https://urn.nsk.hr/urn:nbn:hr:151:866118

Metadata
TitleA comparison of accuracy of image recognition apps for identification of edible weed species
Title (croatian)Usporedba točnosti aplikacija za prepoznavanje slika za identifikaciju jestivih korovnih vrsta
AuthorDomagoj Grgić
MentorMarija Ravlić (mentor)
Committee memberVlatka Rozman (predsjednik povjerenstva)
Committee memberMonika Marković (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Agrobiotechnical Sciences Osijek
(Department of Phytomedicine)
Osijek
Defense date and country2023-09-28, Croatia
Scientific / art field,
discipline and subdiscipline		BIOTECHNICAL SCIENCES
Agronomy
Ecology and Environmental Protection
Abstract
The aim of the research was to determine whether mobile plant identification applications can recognize edible weed species, and to compare and determine which mobile applications yield the most accurate results. For identification purposes, 116 photographs of various parts of 11 plant species found in the city of Osijek were used. The applications used in the research included PictureThis, Pl@ntNet ntNet, and PlantSnap, which are available on Google Play, the App Store, and the internet. PictureThis achieved outstanding results with a 96.55% accuracy rate in plant recognition, followed by Pl@ntNet at 52.59% in the second place. PlantSnap had the lowest accuracy rate, with 50.86% recognition accuracy. Digital plant identification applications can assist in identifying edible weeds; however, in cases of uncertainty, it is important to verify the identification to avoid confusing edible plants with potentially harmful ones. Furthermore, it's worth noting that both PictureThis and PlantSnap require an annual subscription with a seven-day trial period, while Pl@ntNet is available for free.
Abstract (croatian)
Cilj istraživanja je bio utvrditi mogu li mobilne aplikacije za identifikaciju biljaka prepoznati jestive korovne vrste, te usporediti i odrediti koje mobilne aplikacije pokazuju najtočnije rezultate. Za identifikaciju je korišteno 116 fotografija raznih dijelova 11 biljnih vrsta fotografiranih na području grada Osijeka. Aplikacije korištene u istraživanju uključivale su PictureThis, Pl@ntNet te PlantSnap dostupne na Google Play, App store-u i Internetu. PictureThis je postigla vrhunski rezultat od 96,55 % u prepoznavanja biljaka, dok je na drugom mjestu Pl@ntNet s 52,59 % prepoznatih slika. PlantSnap imao je najslabiji rezultat od 50,86 % točnosti prepoznavanja. Digitalne aplikacije za identifikaciju mogu pomoći u prepoznavanju jestivih korova, ali kad postoji nesigurnost, važno je provjeriti identifikaciju kako bi se izbjegla zamjena jestivih biljaka s onima koje su potencijalno štetne. Također, PictureThis i PlantSnap imaju godišnju pretplatu, sa sedmodnevnim probnim periodom dok je Pl@ntNet besplatna.
Keywords
Keywords (croatian)
Languageenglish
URN:NBNurn:nbn:hr:151:866118
Study programmeTitle: Digital Agriculture - graduate - university
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka digitalne poljoprivrede (magistar/magistra inženjer/inženjerka digitalne poljoprivrede)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Public notegraduate thesis
Created on2023-11-16 10:49:34