为开放词汇语义分割中的惰性视觉定位辩护
In Defense of Lazy Visual Grounding for Open-Vocabulary Semantic Segmentation
AI总结：
针对开放词汇语义分割，本文提出无需训练的两阶段 lazy visual grounding，先用迭代 Normalized cuts 发现对象掩码，再迟交互分配文本，并在五个公开数据集上验证其精确定位能力。
摘要：
我们提出了 lazy visual grounding（惰性视觉定位），这是一种先进行无监督对象掩码发现、随后进行对象定位的两阶段方法，用于开放词汇语义分割。以往许多研究将该任务视为缺乏对象级理解的像素到文本分类，利用预训练视觉与语言模型的图像到文本分类能力。我们认为，由于分割本质上是一项视觉任务，视觉对象无需先验文本信息即可区分。惰性视觉定位首先通过迭代 Normalized cuts 发现覆盖整幅图像的对象掩码，随后以迟交互方式为发现的对象分配文本。我们的模型无需额外训练，却在五个公开数据集上表现出色：Pascal VOC、Pascal Context、COCO-object、COCO-stuff 和 ADE 20K。特别是，视觉上富有吸引力的分割结果证明了该模型精确定位对象的能力。论文主页：https://cvlab.postech.ac.kr/research/lazygrounding
英文摘要：
We present lazy visual grounding, a two-stage approach of unsupervised object mask discovery followed by object grounding, for open-vocabulary semantic segmentation. Plenty of the previous art casts this task as pixel-to-text classification without object-level comprehension, leveraging the image-to-text classification capability of pretrained vision-and-language models. We argue that visual objects are distinguishable without the prior text information as segmentation is essentially a vision task. Lazy visual grounding first discovers object masks covering an image with iterative Normalized cuts and then later assigns text on the discovered objects in a late interaction manner. Our model requires no additional training yet shows great performance on five public datasets: Pascal VOC, Pascal Context, COCO-object, COCO-stuff, and ADE 20K. Especially, the visually appealing segmentation results demonstrate the model capability to localize objects precisely. Paper homepage: https://cvlab.postech.ac.kr/research/lazygrounding
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。
