Cityscapes-Panoptic-Parts, PASCAL-Panoptic-Parts
收藏资源简介：
本研究介绍了两个创新数据集：Cityscapes-Panoptic-Parts和PASCAL-Panoptic-Parts，旨在提升图像场景理解能力。这两个数据集均支持全景分割标注，并针对特定语义类别提供部分级别标签。Cityscapes-Panoptic-Parts扩展自Cityscapes数据集，增加了23个部分级别类别的手动标注，而PASCAL-Panoptic-Parts则是通过合并PASCAL-Parts和PASCAL-Context数据集创建，涵盖了80个语义类别和20个实例类别。这些数据集通过统一的标注协议和格式，支持多级图像理解任务，如语义分割、实例分割和对象检测，适用于解决复杂环境中的对象识别和解析问题。
This study introduces two novel datasets: Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts, which are designed to enhance image scene understanding capabilities. Both datasets support panoptic segmentation annotation and provide part-level labels for specific semantic categories. Cityscapes-Panoptic-Parts is extended from the Cityscapes dataset, with manual annotations added for 23 part-level categories. In contrast, PASCAL-Panoptic-Parts is created by merging the PASCAL-Parts and PASCAL-Context datasets, covering 80 semantic categories and 20 instance categories. These datasets adopt a unified annotation protocol and format, supporting multi-level image understanding tasks such as semantic segmentation, instance segmentation, and object detection, and are applicable to solving object recognition and parsing problems in complex environments.
数据集概述
数据集介绍
数据集名称
- Cityscapes-Panoptic-Parts
- PASCAL-Panoptic-Parts
数据集来源
这两个数据集是通过扩展两个已建立的图像场景理解数据集创建的：
数据集描述
详细的描述和各种统计数据在技术报告中提供，链接如下：
数据集下载链接
数据集示例
 |
 |
|---|---|
 |
 |
更多示例请参见这里。
评估指标
我们提供两种评估Panoptic Parts数据集性能的指标：
- Part-aware Panoptic Quality (PartPQ)：详细信息
- Intersection over Union (IoU)：待定（TBA）
引用
如果您发现我们的工作有用或在您的研究中使用，请引用我们：
bibtex @inproceedings{degeus2021panopticparts, title = {Part-aware Panoptic Segmentation}, author = {Daan de Geus and Panagiotis Meletis and Chenyang Lu and Xiaoxiao Wen and Gijs Dubbelman}, booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, year = {2021} }
bibtex @article{meletis2020panopticparts, title = {Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts datasets for Scene Understanding}, author = {Panagiotis Meletis and Xiaoxiao Wen and Chenyang Lu and Daan de Geus and Gijs Dubbelman}, type = {Technical report}, institution = {Eindhoven University of Technology}, date = {16/04/2020}, url = {https://github.com/tue-mps/panoptic_parts}, eprint={2004.07944}, archivePrefix={arXiv}, primaryClass={cs.CV} }
- 1Cityscapes-Panoptic-Parts and PASCAL-Panoptic-Parts datasets for Scene Understanding埃因霍温理工大学 阿姆斯特丹大学 · 2020年
