Multi-domain semantic segmentation with overlapping labels (CROSBI ID 715741)
Prilog sa skupa u zborniku | izvorni znanstveni rad | međunarodna recenzija
Podaci o odgovornosti
Bevandić, Petra ; Oršić, Marin ; Grubišić, Ivan ; Šarić, Josip ; Šegvić, Sinisa
engleski
Multi-domain semantic segmentation with overlapping labels
Deep supervised models have an unprecedented capacity to absorb large quantities of training data. Hence, training on many datasets becomes a method of choice towards graceful degradation in unusual scenes. Unfortunately, different datasets often use incompatible labels. For instance, the Cityscapes road class subsumes all driving surfaces, while Vistas defines separate classes for road markings, manholes etc. We address this challenge by proposing a principled method for seamless learning on datasets with overlapping classes based on partial labels and probabilistic loss. Our method achieves competitive within- dataset and cross-dataset generalization, as well as ability to learn visual concepts which are not separately labeled in any of the training datasets. Experiments reveal competitive or state- of-the-art performance on two multi-domain dataset collections and on the WildDash 2 benchmark.
računalni vid ; semantička segmentacija
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
2422-2431.
2022.
objavljeno
10.1109/wacv51458.2022.00248
Podaci o matičnoj publikaciji
Proceeedings of IEEE/CVF Winter Conference on Applications of Computer Vision
Bowyer, Kevin ; Medioni, Gérard ; Scheirer, Walter
Waikoloa (HI): Institute of Electrical and Electronics Engineers (IEEE)
978-1-6654-0915-5
Podaci o skupu
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
poster
03.01.2022-08.01.2022
Waikoloa (HI), Sjedinjene Američke Države