arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2006.02158cs.CV
收藏

基于插值的半监督学习的目标检测

Interpolation-based semi-supervised learning for object detection

Seoul National University(首尔国立大学);Aalto University(阿尔托大学);Mila Université de Montréal(蒙特利尔大学米拉研究所);SK hynix(SK海力士)

Jisoo Jeong, Vikas Verma, Minsung Hyun, Juho Kannala, Nojun Kwak

摘要:

尽管目标检测任务的数据标记成本远高于分类任务,但用于目标检测的半监督学习方法还没有得到太多的研究。本文提出了一种基于插值的半监督学习目标检测方法(ISD),该方法考虑并解决了传统的插值正则化(IR)直接应用于目标检测所带来的问题。我们根据IR中混合的两个原始补丁的对象性得分将模型的输出分为两种类型。然后,我们以无监督的方式应用适合于每种类型的单独损失。提出的损失显着提高性能的半监督学习以及监督学习。在监督学习设置中,我们的方法显著地改进了基线方法。在半监督学习环境中,我们的算法提高了性能的基准数据集(PASCAL VOC和MSCOCO)在基准架构(SSD)。

英文摘要:

Despite the data labeling cost for the object detection tasks being substantially more than that of the classification tasks, semi-supervised learning methods for object detection have not been studied much. In this paper, we propose an Interpolation-based Semi-supervised learning method for object Detection (ISD), which considers and solves the problems caused by applying conventional Interpolation Regularization (IR) directly to object detection. We divide the output of the model into two types according to the objectness scores of both original patches that are mixed in IR. Then, we apply a separate loss suitable for each type in an unsupervised manner. The proposed losses dramatically improve the performance of semi-supervised learning as well as supervised learning. In the supervised learning setting, our method improves the baseline methods by a significant margin. In the semi-supervised learning setting, our algorithm improves the performance on a benchmark dataset (PASCAL VOC and MSCOCO) in a benchmark architecture (SSD).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑