Learning with Free Object Segments for Long-Tailed Instance Segmentation
Learning with Free Object Segments for Long-Tailed Instance Segmentation
复制标题
DOI：
10.1007/978-3-031-20080-9_38
复制
发表时间：
2022-02
期刊：
影响因子：
--
通讯作者：
Cheng Zhang;Tai-Yu Pan;Tianle Chen;Jike Zhong;Wen-juan Fu;Wei-Lun Chao
中科院分区：
文献类型：
--
作者：
Cheng Zhang;Tai-Yu Pan;Tianle Chen;Jike Zhong;Wen-juan Fu;Wei-Lun Chao
为复杂场景中的大量类构建实例分割模型的一个基本挑战是缺乏训练示例，特别是对于稀有对象。在本文中，我们探索了增加训练示例而无需费力的数据收集和注释的可能性。我们发现，大量的实例段可以从以对象为中心的图像中自由地获得，根据两个见解：（i）以对象为中心的图像通常包含一个简单背景中的突出对象;（ii）来自同一类的对象通常具有相似的外观或与背景相似的对比度。受这些见解的启发，我们提出了一个简单的和可扩展的frameworkFreeSegfor提取和利用这些“自由”对象前景段，以促进模型训练长尾实例分割。具体地说，我们调查了同一类的以对象为中心的图像之间的相似性，以提出前景实例的候选片段，然后对片段质量进行新的排名。然后可以使用所得到的高质量对象片段来增强现有的长尾数据集，例如，通过将片段复制并粘贴到原始训练图像上。大量的实验表明，FreeSegment在强基线的基础上取得了实质性的改进，并在分割稀有对象类别方面达到了最先进的准确性。我们的代码可在https://github.com/czhang0528/FreeSeg上公开获取。
One fundamental challenge in building an instance segmentation model for a large number of classes in complex scenes is the lack of training examples, especially for rare objects. In this paper, we explore the possibility to increase the training examples without laborious data collection and annotation. We find that an abundance of instance segments can potentially be obtained freely from object-centric images, according to two insights: (i) an object-centric image usually contains one salient object in a simple background; (ii) objects from the same class often share similar appearances or similar contrasts to the background. Motivated by these insights, we propose a simple and scalable frameworkFreeSegfor extracting and leveraging these “free” object foreground segments to facilitate model training in long-tailed instance segmentation. Concretely, we investigate the similarity among object-centric images of the same class to propose candidate segments of foreground instances, followed by a novel ranking of segment quality. The resulting high-quality object segments can then be used to augment the existing long-tailed datasets,e.g., by copying and pasting the segments onto the original training images. Extensive experiments show thatFreeSegyields substantial improvements on top of strong baselines and achieves state-of-the-art accuracy for segmenting rare object categories. Our code is publicly available at https://github.com/czhang0528/FreeSeg.