IoU losses are surrogates that directly optimize the Jaccard index. In
semantic segmentation, IoU losses are shown to perform better with respect to
the Jaccard index measure than pixel-wise losses such as the cross-entropy
loss. The most notable IoU losses are the soft Jaccard loss and the
Lovasz-Softmax loss. However, these losses are incompatible with soft labels
which are ubiquitous in machine learning. In this paper, we propose Jaccard
metric losses (JMLs), which are variants of the soft Jaccard loss, and are
compatible with soft labels. With JMLs, we study two of the most popular use
cases of soft labels: label smoothing and knowledge distillation. With a
variety of architectures, our experiments show significant improvements over
the cross-entropy loss on three semantic segmentation datasets (Cityscapes,
PASCAL VOC and DeepGlobe Land), and our simple approach outperforms
state-of-the-art knowledge distillation methods by a large margin. Our source
code is available at:
\href{this https URL}{this https URL}.

本文提出了一种称为 Jaccard metric losses（JMLs）的 IoU loss 变体，可与软标签兼容，在 Cityscapes、PASCAL VOC 和 DeepGlobe Land 三个语义分割数据集上相对于交叉熵损失有显着的改进，并且在知识蒸馏方面胜过了现有的最优方法。