探索面向统一且高效图像抠图的交互式引导
Exploring the Interactive Guidance for Unified and Effective Image Matting
AI总结：
本文提出统一交互式图像抠图方法 UIM，通过多类型交互确定目标，并将抠图解耦为前景分割与透明度预测，结合多尺度注意力融合提升透明及不透明物体抠图效果。
摘要：
近期的图像抠图研究正朝着提出无 trimap（三分图）或交互式方法的方向发展，以完成复杂的图像抠图任务。尽管避免了 trimap 标注的大量人力，现有方法仍存在两个局限：（1）对于包含多个对象的单幅图像，必须提供额外的交互信息以帮助确定抠图目标；（2）对于透明对象，与不透明对象相比，从 RGB 图像准确回归 alpha matte（透明度遮罩）要困难得多。在本工作中，我们提出了一种统一交互式图像抠图方法，名为 UIM，它解决了上述局限，并在任意场景下取得了令人满意的抠图结果。具体而言，UIM 利用多种类型的用户交互来避免多个抠图目标的歧义，并且我们详细比较了不同标注类型的优缺点。为统一透明与不透明对象的抠图性能，我们将图像抠图解耦为两个阶段，即前景分割和透明度预测。此外，我们设计了一个多尺度注意力融合模块，以缓解边界区域的模糊问题。实验结果表明，UIM 在 Composition-1K 测试集和一个合成统一数据集上取得了最先进的性能。
英文摘要：
Recent image matting studies are developing towards proposing trimap-free or interactive methods for complete complex image matting tasks. Although avoiding the extensive labors of trimap annotation, existing methods still suffer from two limitations: (1) For the single image with multiple objects, it is essential to provide extra interaction information to help determining the matting target; (2) For transparent objects, the accurate regression of alpha matte from RGB image is much more difficult compared with the opaque ones. In this work, we propose a Unified Interactive image Matting method, named UIM, which solves the limitations and achieves satisfying matting results for any scenario. Specifically, UIM leverages multiple types of user interaction to avoid the ambiguity of multiple matting targets, and we compare the pros and cons of different annotation types in detail. To unify the matting performance for transparent and opaque objects, we decouple image matting into two stages, i.e., foreground segmentation and transparency prediction. Moreover, we design a multi-scale attentive fusion module to alleviate the vagueness in the boundary region. Experimental results demonstrate that UIM achieves state-of-the-art performance on the Composition-1K test set and a synthetic unified dataset.
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。
