PCLoss: Fashion Landmark Estimation with Position Constraint Loss
Lists
Only you see this
Abstract
Highlights•Help network search landmark positions in a more reasonable region•Alleviate outliers and duplicate detection problems in fashion landmark estimation by loss function•Easily applied to many popular CNN models without extra computation during inference•Introduce the skeleton-like characteristic of fashion landmarks to strength the position constraint
BibTeX
@article{song2021pcloss,
author = {Meijia Song and Hong Liu and Wei Shi and Xia Li},
title = {{PCLoss: Fashion Landmark Estimation with Position Constraint Loss}},
journal = {Pattern Recognition},
year = {2021},
doi = {10.1016/j.patcog.2021.108028},
}Loading related content...