The task of semantic segmentation is more complex than the task of image classification and object search, which is due not only to the need to determine the classes of objects, but also to identify their structure, the correct allocation of parts of objects in the image.
All images in the dataset are presented in high quality (at least 1800 pixels in height and more than 1200 in width). The most variable and diverse sets of clothes have been selected. The models stand straight up to their full height. The background is separated.
Color values of masks in rgb:
Men's topwear: (236,233,0)
Women's topwear (blouses, blouses, etc.): (221,75,75)
Outerwear (jackets, raincoats, coats): (124,88,18)
Button-down vest: (128,34,155)
Men's Sweater/Hoodie: (24,45,41)
The dataset can be used for categorization tasks, training neural networks in color matching, creating visual models, and so on. The resulting neurons can be used in such areas as fashion retail, augmented reality, ecommerce.