arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.01418cs.CV
收藏

一种用于无需训练的类别无关计数的简单而有效基线

A Simple-but-effective Baseline for Training-free Class-Agnostic Counting

Australian Institution for Machine Learning(澳大利亚机器学习研究院);The University of Adelaide(阿德莱德大学)

Yuhao Lin, Haiming Xu, Lingqiao Liu, Javen Qinfeng Shi

AI总结:

针对无需训练的类别无关计数性能不足问题,本文提出结合超像素提示、语义编码器、多尺度机制与直推式原型的简单基线,显著提升效果并追平训练型方法。

摘要:

类别无关计数(Class-Agnostic Counting, CAC)旨在仅借助少量参考示例,准确计数给定图像中的物体。以往实现该目标的方法依赖额外训练,而近期研究表明,可以利用既有的基础模型,尤其是 Segment Anything Model(SAM),通过实例级分割进行计数,从而在无需训练的情况下完成该任务。尽管前景良好,但当前无需训练的方法在性能上仍落后于基于训练的同类方法。在本研究中,我们提出一种直接的无需训练方案,可有效弥合这一性能差距,并作为一个强有力的基线。我们工作的主要贡献在于发现了四项能够提升性能的关键技术。具体而言,我们建议采用超像素算法生成更精确的初始点提示,利用具备更丰富语义知识的图像编码器替代 SAM 编码器来表示候选对象,并采用多尺度机制和直推式原型方案来更新参考示例的表示。通过结合这四项技术,我们的方法相较现有无需训练的方法取得了显著提升,并达到了与基于训练方法相当的性能。

英文摘要:

Class-Agnostic Counting (CAC) seeks to accurately count objects in a given image with only a few reference examples. While previous methods achieving this relied on additional training, recent efforts have shown that it's possible to accomplish this without training by utilizing pre-existing foundation models, particularly the Segment Anything Model (SAM), for counting via instance-level segmentation. Although promising, current training-free methods still lag behind their training-based counterparts in terms of performance. In this research, we present a straightforward training-free solution that effectively bridges this performance gap, serving as a strong baseline. The primary contribution of our work lies in the discovery of four key technologies that can enhance performance. Specifically, we suggest employing a superpixel algorithm to generate more precise initial point prompts, utilizing an image encoder with richer semantic knowledge to replace the SAM encoder for representing candidate objects, and adopting a multiscale mechanism and a transductive prototype scheme to update the representation of reference examples. By combining these four technologies, our approach achieves significant improvements over existing training-free methods and delivers performance on par with training-based ones.

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。

↑