Few-Shot Pattern Detection via Template Matching and Regression

TL;DR

This paper introduces TMR, a simple and effective few-shot pattern detection method using template matching and regression, which outperforms existing approaches and generalizes well across diverse datasets.

Contribution

The paper proposes TMR, a novel few-shot pattern detection approach that preserves spatial information and leverages classic template matching, along with a new diverse dataset RPINE.

Findings

TMR outperforms state-of-the-art methods on three benchmarks.

TMR demonstrates strong cross-dataset generalization.

RPINE dataset covers a wider range of patterns than existing datasets.

Abstract

We address the problem of few-shot pattern detection, which aims to detect all instances of a given pattern, typically represented by a few exemplars, from an input image. Although similar problems have been studied in few-shot object counting and detection (FSCD), previous methods and their benchmarks have narrowed patterns of interest to object categories and often fail to localize non-object patterns. In this work, we propose a simple yet effective detector based on template matching and regression, dubbed TMR. While previous FSCD methods typically represent target exemplars as spatially collapsed prototypes and lose structural information, we revisit classic template matching and regression. It effectively preserves and leverages the spatial layout of exemplars through a minimalistic structure with a small number of learnable convolutional or projection layers on top of a frozen backbone We also introduce a new dataset, dubbed RPINE, which covers a wider range of patterns than existing object-centric datasets. Our method outperforms the state-of-the-art methods on the three benchmarks, RPINE, FSCD-147, and FSCD-LVIS, and demonstrates strong generalization in cross-dataset evaluation.