The Ciona17 dataset for semantic segmentation of invasive species in a marine aquaculture environment

An original dataset for semantic segmentation, Ciona17, is introduced, which to the best of the authors' knowledge, is the first dataset of its kind with pixel-level annotations pertaining to invasive species in a marine environment. Diverse outdoor illumination, a range of object shapes, and severe occlusion provide a significant challenge for the computer vision community. An accompanying ground-truthing tool for superpixel labeling, Truth and Crop, is also introduced. In a subsequent work, results are reported in terms of the mean intersection over union (mIoU) with segmentation mask. The GUI application for ground-truthing semantic segmentation datasets in PyQt4/OpenCV can be accessed at https://github.com/AngusG/truth-and-crop

本文介绍了一款用于语义分割的原创数据集Ciona17。据作者所知，该数据集是首个针对海洋环境入侵物种的像素级标注语义分割数据集。该数据集涵盖多样的户外光照条件、各异的目标形态以及严重的遮挡场景，为计算机视觉领域带来了显著挑战。本文同时推出了一款用于超像素标注的真值标注工具Truth and Crop。在后续相关研究中，已有基于分割掩码的平均交并比（mean intersection over union，mIoU）指标的实验结果被报道。该基于PyQt4/OpenCV开发的语义分割数据集真值标注GUI应用，可通过以下链接获取：https://github.com/AngusG/truth-and-crop