arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.05095cs.CV
收藏

ST ++:使自训练更好地用于半监督语义分割

ST++: Make Self-training Work Better for Semi-supervised Semantic Segmentation

Nanjing University(南京大学);Tencent(腾讯);Southeast University(东南大学)

Lihe Yang, Wei Zhuo, Lei Qi, Yinghuan Shi, Yang Gao

摘要:

通过伪标签进行自我训练是一种传统、简单和流行的管道,可以利用未标记的数据。在这项工作中,我们首先构建了一个强大的基线的自我训练(即ST)的半监督语义分割通过注入强数据增强(SDA)对未标记的图像,以减轻过度拟合嘈杂的标签,以及解耦教师和学生之间的类似的预测。有了这个简单的机制,我们的ST优于所有现有的方法,没有任何花里胡哨的东西,例如,迭代再训练。受这些令人印象深刻的结果的启发,我们深入研究了SDA,并提供了一些实证分析。然而,不正确的伪标签仍然容易积累并降低性能。为此,我们进一步提出了一个先进的自训练框架(即ST++),它通过基于整体预测级稳定性优先考虑可靠的未标记图像来执行选择性再训练。具体来说,在第一阶段的监督训练中保存了几个模型检查点,它们对未标记图像的预测差异作为可靠性的衡量标准。我们的图像级选择为学习提供了全面的上下文信息。我们证明,它是更适合分割比常见的像素明智的选择。因此,ST++进一步提升了ST的性能。代码可在https://github.com/LiheYoung/ST-PlusPlus上找到。

英文摘要:

Self-training via pseudo labeling is a conventional, simple, and popular pipeline to leverage unlabeled data. In this work, we first construct a strong baseline of self-training (namely ST) for semi-supervised semantic segmentation via injecting strong data augmentations (SDA) on unlabeled images to alleviate overfitting noisy labels as well as decouple similar predictions between the teacher and student. With this simple mechanism, our ST outperforms all existing methods without any bells and whistles, e.g., iterative re-training. Inspired by the impressive results, we thoroughly investigate the SDA and provide some empirical analysis. Nevertheless, incorrect pseudo labels are still prone to accumulate and degrade the performance. To this end, we further propose an advanced self-training framework (namely ST++), that performs selective re-training via prioritizing reliable unlabeled images based on holistic prediction-level stability. Concretely, several model checkpoints are saved in the first stage supervised training, and the discrepancy of their predictions on the unlabeled image serves as a measurement for reliability. Our image-level selection offers holistic contextual information for learning. We demonstrate that it is more suitable for segmentation than common pixel-wise selection. As a result, ST++ further boosts the performance of our ST. Code is available at https://github.com/LiheYoung/ST-PlusPlus.

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。

↑