SPIN: Hierarchical Segmentation with Subpart Granularity in Natural Images
This work addresses the need for hierarchical segmentation with subpart granularity in natural images, providing a new dataset and metrics for the computer vision community, though it is incremental in building upon existing segmentation frameworks.
The authors introduced SPIN, the first hierarchical semantic segmentation dataset with subpart annotations for natural images, and proposed two novel evaluation metrics to assess algorithms' ability to capture spatial and semantic relationships across hierarchical levels. They benchmarked modern models across three tasks and publicly released the dataset to enable community-wide progress.
Hierarchical segmentation entails creating segmentations at varying levels of granularity. We introduce the first hierarchical semantic segmentation dataset with subpart annotations for natural images, which we call SPIN (SubPartImageNet). We also introduce two novel evaluation metrics to evaluate how well algorithms capture spatial and semantic relationships across hierarchical levels. We benchmark modern models across three different tasks and analyze their strengths and weaknesses across objects, parts, and subparts. To facilitate community-wide progress, we publicly release our dataset at https://joshmyersdean.github.io/spin/index.html.