The semantic segmentation of clothes is a challenging task due to the wide variety of clothing styles, layers and shapes.
The UTFPR-SBD3 contains 4,500 images manually annotated at pixel level in 18 classes plus background.
To ensure the high quality of the dataset, all images were manually annotated at the pixel level using JS Segment Annotator, 2 a free web-based image annotation tool. The raw images were carefully selected to avoid, as far as possible, classes with low number of instances.
Variants: UTFPR-SBD3
This dataset is used in 1 benchmark:
No recent benchmark submissions available for this dataset.
No papers with results on this dataset found.