Single-Image Localised Reflection Removal with k-Order Differences Term
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F25%3A00386497" target="_blank" >RIV/68407700:21230/25:00386497 - isvavai.cz</a>
Result on the web
<a href="https://doi.org/10.1007/978-3-031-95918-9_8" target="_blank" >https://doi.org/10.1007/978-3-031-95918-9_8</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1007/978-3-031-95918-9_8" target="_blank" >10.1007/978-3-031-95918-9_8</a>
Alternative languages
Result language
angličtina
Original language name
Single-Image Localised Reflection Removal with k-Order Differences Term
Original language description
We introduce the problem of localized glass-reflection removal (LGRR), which targets the removal of highlights caused by light reflections on glass surfaces. Our approach uses an end-to-end convolutional neural network trained on MS-COCO image crops with synthetic reflections generated from real light source images. We propose a cost function that includes image difference terms and show that it improves reflection removal and inpainting compared to standard L1 and L2 losses. Experimental results demonstrate that our method effectively reduces localized reflections.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
Result continuities
Project
<a href="/en/project/GA25-15993S" target="_blank" >GA25-15993S: Detection of anomalous or out-of-distribution inputs in deep neural network models</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)
Others
Publication year
2025
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
Image Analysis
ISBN
978-3-031-95918-9
ISSN
0302-9743
e-ISSN
1611-3349
Number of pages
14
Pages from-to
106-119
Publisher name
Springer, Cham
Place of publication
—
Event location
Reykjavik
Event date
Jun 23, 2025
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
001553877800008