超越背景偏移：重新思考持续语义分割中的实例回放
Beyond Background Shift: Rethinking Instance Replay in Continual Semantic Segmentation
AI总结：
本文针对持续语义分割中存储图像导致的背景混淆问题，提出增强实例回放（EIR）方法，通过实例存储与整合缓解背景偏移，显著超越现有最先进方法。
摘要：
本文聚焦于持续语义分割（CSS），该任务要求分割网络在不遗忘先前学习类别知识的情况下，持续学习新类别。尽管存储旧类别图像并将其直接纳入新模型训练已被证明能有效缓解分类任务中的灾难性遗忘，但这一策略在CSS中表现出显著局限性。具体而言，存储图像与新图像带有部分类别标注，导致未标注类别与背景之间产生混淆，使模型拟合复杂化。为解决此问题，本文提出一种新颖的增强实例回放（EIR）方法，该方法不仅通过旧类别的实例存储保留旧类别知识，同时消除背景混淆，还通过将存储实例与新图像整合来缓解新图像中的背景偏移。通过有效解决存储图像和新图像中的背景偏移，EIR缓解了CSS任务中的灾难性遗忘，从而增强了模型在CSS上的能力。实验结果验证了本方法的有效性，其显著优于最先进的CSS方法。
英文摘要：
In this work, we focus on continual semantic segmentation (CSS), where segmentation networks are required to continuously learn new classes without erasing knowledge of previously learned ones. Although storing images of old classes and directly incorporating them into the training of new models has proven effective in mitigating catastrophic forgetting in classification tasks, this strategy presents notable limitations in CSS. Specifically, the stored and new images with partial category annotations leads to confusion between unannotated categories and the background, complicating model fitting. To tackle this issue, this paper proposes a novel Enhanced Instance Replay (EIR) method, which not only preserves knowledge of old classes while simultaneously eliminating background confusion by instance storage of old classes, but also mitigates background shifts in the new images by integrating stored instances with new images. By effectively resolving background shifts in both stored and new images, EIR alleviates catastrophic forgetting in the CSS task, thereby enhancing the model's capacity for CSS. Experimental results validate the efficacy of our approach, which significantly outperforms state-of-the-art CSS methods.
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。
