BriefGPT.xyz
Jul, 2023
语义分割中的标签校准在领域转换下
Label Calibration for Semantic Segmentation Under Domain Shift
HTML
PDF
Ondrej Bohdal, Da Li, Timothy Hospedales
TL;DR
预训练模型可通过计算域迁移下的软标签原型并根据与预测类别概率最接近的原型进行预测,从而适应无标签目标域数据,这种适应过程快速且几乎不需要计算资源,且能显著提升性能,我们在实用的合成到真实场景的语义分割问题中证明了该标签校准的益处。
Abstract
Performance of a pre-trained
semantic segmentation
model is likely to substantially decrease on data from a new domain. We show a
pre-trained model
can be adapted to unlabelled target domain data by calculating <
→