论文标题
HCFORMER：统一图像分割和分层群集
HCFormer: Unified Image Segmentation with Hierarchical Clustering
论文作者
论文摘要
分层聚类是一种广泛用于经典图像分割方法的有效和有效方法。但是，许多使用神经网络的现有方法直接从每个像素功能中生成分割掩码，从而使体系结构设计变得复杂并降低了解释性。在这项工作中，我们提出了一个更简单，更容易解释的架构，称为HCFormer。 HCFormer通过自下而上的分层聚类来完成图像分割，并使我们能够将中间结果解释，可视化和评估为分层聚类结果。 HCFORMER可以使用相同的体系结构来解决语义，实例和全盘分段，因为像素聚类是各种图像分割任务的常见方法。在实验中，与语义分割（ADE20K上的55.5 MIOU），实例分割（可可二的47.1 AP）和泛型分段（可可在Coco上为55.7 PQ）相比，HCFormer达到了可比或卓越的分割精度。
Hierarchical clustering is an effective and efficient approach widely used for classical image segmentation methods. However, many existing methods using neural networks generate segmentation masks directly from per-pixel features, complicating the architecture design and degrading the interpretability. In this work, we propose a simpler, more interpretable architecture, called HCFormer. HCFormer accomplishes image segmentation by bottom-up hierarchical clustering and allows us to interpret, visualize, and evaluate the intermediate results as hierarchical clustering results. HCFormer can address semantic, instance, and panoptic segmentation with the same architecture because the pixel clustering is a common approach for various image segmentation tasks. In experiments, HCFormer achieves comparable or superior segmentation accuracy compared to baseline methods on semantic segmentation (55.5 mIoU on ADE20K), instance segmentation (47.1 AP on COCO), and panoptic segmentation (55.7 PQ on COCO).
