arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.02001cs.CV
收藏

零样本目标计数

Zero-shot Object Counting

Stony Brook University(石溪大学);EPFL(洛桑联邦理工学院);Amazon(亚马逊)

Jingyi Xu, Hieu Le, Vu Nguyen, Viresh Ranjan, Dimitris Samaras

AI总结:

提出零样本目标计数设定 ZSC,仅凭类别名称自动筛选最佳图像块作为计数样本,无需人工标注,并在 FSC-147 上验证有效。

摘要:

类别无关的目标计数旨在测试时统计任意类别的目标实例数量。这一任务颇具挑战,但也能支持许多潜在应用。现有方法需要人工标注的样本作为输入,而对于新类别而言这类样本往往无法获得,尤其是在自主系统中。因此,我们提出零样本目标计数(ZSC)这一新设定:测试时仅有类别名称可用。这样的计数系统不需要人工标注者参与流程,能够自动运行。从类别名称出发,我们提出一种可准确识别最佳图像块的方法,这些图像块随后可作为计数样本。具体而言,我们首先构建类别原型,用于选择可能包含感兴趣目标的图像块,即类别相关图像块。此外,我们引入一个模型,能够定量衡量任意图像块作为计数样本的适合程度。将该模型应用于所有候选图像块后,我们可以选出最合适的图像块作为计数样本。在近期的类别无关计数数据集 FSC-147 上的实验结果验证了本方法的有效性。代码位于 https://github.com/cvlab-stonybrook/zero-shot-counting。

英文摘要:

Class-agnostic object counting aims to count object instances of an arbitrary class at test time. It is challenging but also enables many potential applications. Current methods require human-annotated exemplars as inputs which are often unavailable for novel categories, especially for autonomous systems. Thus, we propose zero-shot object counting (ZSC), a new setting where only the class name is available during test time. Such a counting system does not require human annotators in the loop and can operate automatically. Starting from a class name, we propose a method that can accurately identify the optimal patches which can then be used as counting exemplars. Specifically, we first construct a class prototype to select the patches that are likely to contain the objects of interest, namely class-relevant patches. Furthermore, we introduce a model that can quantitatively measure how suitable an arbitrary patch is as a counting exemplar. By applying this model to all the candidate patches, we can select the most suitable patches as exemplars for counting. Experimental results on a recent class-agnostic counting dataset, FSC-147, validate the effectiveness of our method. Code is available at https://github.com/cvlab-stonybrook/zero-shot-counting

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。

↑