Matches in SemOpenAlex for { <https://semopenalex.org/work/W3083179931> ?p ?o ?g. }
Showing items 1 to 64 of
64
with 100 items per page.
- W3083179931 abstract "Robust segmentation of thermal hand images, a trivial task in controlled environments, can be difficult for unconstrained acquisition with a thermal camera, when the temperature of the object differs little from this of a background. This paper implements a method for segmenting hand images collected in the thermal spectrum involving a pre-trained, off-the-shelf data-driven model, fine-tuned with multiple databases of thermal hand images and corresponding, manually annotated ground truth masks. This allows superior performance for difficult samples, compared against conventional methods such as Otsu’s thresholding and Gaussian Mixture Modeling. The segmentation accuracy for good quality samples is comparable with these traditional methods, while at the same time being more unconstrained acquisition-resistant. An improvement of up to 20% in accuracy measured by intersection over union is observed for difficult samples, such as hands that were partially colder than the background, as well as those with wristwatches and jewelry. This robust segmentation in various acquisition scenarios allows not only the correct localization of regions of interest for feature extraction, but also for extracting accurate hand geometry information. Together with the results of experiments, involving multiple training and testing scenarios on three different databases, we provide codes, model weights, and ground truth masks to ensure reproducibility and facilitate further research." @default.
- W3083179931 created "2020-09-11" @default.
- W3083179931 creator A5035145746 @default.
- W3083179931 creator A5086290876 @default.
- W3083179931 date "2019-09-01" @default.
- W3083179931 modified "2023-09-26" @default.
- W3083179931 title "Unconstrained Thermal Hand Segmentation" @default.
- W3083179931 cites W1979155176 @default.
- W3083179931 cites W2000805497 @default.
- W3083179931 cites W2019320651 @default.
- W3083179931 cites W2024129051 @default.
- W3083179931 cites W2039004170 @default.
- W3083179931 cites W2058668884 @default.
- W3083179931 cites W2073266290 @default.
- W3083179931 cites W2103729193 @default.
- W3083179931 cites W2119416456 @default.
- W3083179931 cites W2122393977 @default.
- W3083179931 cites W2125139591 @default.
- W3083179931 cites W2129915901 @default.
- W3083179931 cites W2133059825 @default.
- W3083179931 cites W2741839910 @default.
- W3083179931 cites W2765685479 @default.
- W3083179931 cites W2783543995 @default.
- W3083179931 cites W2795311882 @default.
- W3083179931 cites W2809810108 @default.
- W3083179931 cites W2889100879 @default.
- W3083179931 cites W2890653641 @default.
- W3083179931 cites W2963881378 @default.
- W3083179931 cites W3006153529 @default.
- W3083179931 cites W3101650140 @default.
- W3083179931 cites W4243694068 @default.
- W3083179931 doi "https://doi.org/10.1109/btas46853.2019.9185976" @default.
- W3083179931 hasPublicationYear "2019" @default.
- W3083179931 type Work @default.
- W3083179931 sameAs 3083179931 @default.
- W3083179931 citedByCount "0" @default.
- W3083179931 crossrefType "proceedings-article" @default.
- W3083179931 hasAuthorship W3083179931A5035145746 @default.
- W3083179931 hasAuthorship W3083179931A5086290876 @default.
- W3083179931 hasConcept C154945302 @default.
- W3083179931 hasConcept C31972630 @default.
- W3083179931 hasConcept C41008148 @default.
- W3083179931 hasConcept C89600930 @default.
- W3083179931 hasConceptScore W3083179931C154945302 @default.
- W3083179931 hasConceptScore W3083179931C31972630 @default.
- W3083179931 hasConceptScore W3083179931C41008148 @default.
- W3083179931 hasConceptScore W3083179931C89600930 @default.
- W3083179931 hasLocation W30831799311 @default.
- W3083179931 hasOpenAccess W3083179931 @default.
- W3083179931 hasPrimaryLocation W30831799311 @default.
- W3083179931 hasRelatedWork W1669643531 @default.
- W3083179931 hasRelatedWork W2005437358 @default.
- W3083179931 hasRelatedWork W2008656436 @default.
- W3083179931 hasRelatedWork W2023558673 @default.
- W3083179931 hasRelatedWork W2039154422 @default.
- W3083179931 hasRelatedWork W2122581818 @default.
- W3083179931 hasRelatedWork W2134924024 @default.
- W3083179931 hasRelatedWork W2517104666 @default.
- W3083179931 hasRelatedWork W2895616727 @default.
- W3083179931 hasRelatedWork W2182382398 @default.
- W3083179931 isParatext "false" @default.
- W3083179931 isRetracted "false" @default.
- W3083179931 magId "3083179931" @default.
- W3083179931 workType "article" @default.