Your browser does not allow JavaScript!
JavaScript is necessary for the proper functioning of this website. Please enable JavaScript or use a modern browser.
Repository of the University of Ljubljana
Open Science Slovenia
Open Science
DiKUL
slv
|
eng
Search
Browse
New in RUL
About RUL
In numbers
Help
Sign in
Details
Klasifikacija slik s pomočjo toplotne karte pogleda
ID
NABERGOJ, JURIJ
(
Author
),
ID
Žabkar, Jure
(
Mentor
)
More about this mentor...
PDF - Presentation file,
Download
(1,74 MB)
MD5: 851602BFEC59C397DC8E785BB6D2E259
Image galllery
Abstract
Splošno namenski slikovni klasifikacijski modeli dosegajo odlične rezultate na zahtevnih podatkovnih množicah. Kljub temu je klasifikacija z malo učnimi slikami in veliko šuma zahtevna. Za praktično uporabnost klasifikacijskih modelov v takšnih okoliščinah potrebujemo dodatne teoretične in empirične vpoglede. V diplomskem delu pokažemo, da lahko klasifikacijsko točnost s šumnimi podatki in majhnimi učnimi množicami izboljšamo z uporabo toplotnih kart pogleda. Primerjamo točnost običajnih modelov s takimi, ki uporabljajo umetno generirane karte pogleda ali karte, ustvarjene s sledilnikom oči. Primerjavo opravimo na sintetičnih in realnih podatkih, kjer opazimo bistveno boljšo klasifikacijsko točnost z vzorcem 10 do 1000 učnih slik. Naši rezultati so konsistentni pri uporabi standardnih ali prednaučenih nevronskih mrež. Naše ugotovitve spodbujajo širšo uporabo kart pogleda za klasifikacijo slik z majhnimi učnimi množicami in močnim šumom.
Language:
Slovenian
Keywords:
klasifikacija slik
,
karte pogleda
,
šumni podatki
,
majhne učne množice
Work type:
Bachelor thesis/paper
Typology:
2.11 - Undergraduate Thesis
Organization:
FRI - Faculty of Computer and Information Science
FMF - Faculty of Mathematics and Physics
Year:
2023
PID:
20.500.12556/RUL-144754
COBISS.SI-ID:
147963139
Publication date in RUL:
10.03.2023
Views:
1074
Downloads:
133
Metadata:
Cite this work
Plain text
BibTeX
EndNote XML
EndNote/Refer
RIS
ABNT
ACM Ref
AMA
APA
Chicago 17th Author-Date
Harvard
IEEE
ISO 690
MLA
Vancouver
:
NABERGOJ, JURIJ, 2023,
Klasifikacija slik s pomočjo toplotne karte pogleda
[online]. Bachelor’s thesis. [Accessed 19 May 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=144754
Copy citation
Share:
Secondary language
Language:
English
Title:
Image classification using saliency maps
Abstract:
General purpose image classification models have shown great performance on challenging benchmarks. However, classification still poses a challenge when dealing with small training sets and noisy images. Practical model usability in such scenarios requires more theoretical and empirical insights. We show that image classification accuracy on noisy data with small training sets can be improved using saliency heatmaps. We compare classification accuracies of models that use synthetic or eye-tracker generated heatmaps and models without them. We perform the comparison on synthetic and real data, where we observe substantially better accuracy with 10 to 1000 training images. Our results are consistent when using standard or pre-trained neural networks. Our findings encourage broader use of sailency maps in small data image classification with heavy noise.
Keywords:
image classification
,
saliency maps
,
noisy data
,
small dataset
Similar documents
Similar works from RUL:
Searching for similar works...
Similar works from other Slovenian collections:
Back