Jewelry segmentation masks for the 11k Hands dataset

We provide an additional set of segmentation masks for jewelry in the 11K Hands dataset [1]. We filtered out a total of 3179 hands 
with jewelry and were manually annotated using CVAT. For ease of use, the maks have the same size and filename as the original images and are exported in png format. The pixel value represents whether jewelry exists, being 0 background and 1 jewelry.

The 11k Hands [1] dataset is a collection of 11,076 hand photos (1600 × 1200 pixels) from 190 people aged 18 to 75 years old. Each hand was shot from both the dorsal and palmar sides, on a uniform white background, at roughly the same distance from the camera. Each image has a metadata record that includes the following information: the subject ID, gender, age, skin color, and a set of information about the captured hand, such as right- or left-hand, hand side (dorsal or palmar), and logical indicators indicating whether the hand image contains accessories, nail polish, or irregularities. You can download here the original 11K Hands dataset and the metadata.

In the future, we will add our paper if accepted. In the meantime, if you use the masks provided on this webpage, please cite our DOI: 10.5281/zenodo.6541286 and the original 11K Hands paper.

[1] Mahmoud Afifi, "11K Hands: Gender recognition and biometric identification using a large dataset of hand images." Multimedia Tools and Applications, 2019.