Doodle Your Keypoints: Sketch-Based Few-Shot Keypoint Detection

TL;DR

This paper introduces a novel sketch-based framework for few-shot keypoint detection that overcomes cross-modal and style variation challenges, enabling effective learning with minimal data.

Contribution

It proposes a prototypical, grid-based domain adaptation method that leverages sketches as a source-free modality for few-shot keypoint detection.

Findings

Effective in few-shot convergence across new keypoints and classes

Achieves robust cross-modal keypoint detection

Demonstrates superior performance through extensive experiments

Abstract

Keypoint detection, integral to modern machine perception, faces challenges in few-shot learning, particularly when source data from the same distribution as the query is unavailable. This gap is addressed by leveraging sketches, a popular form of human expression, providing a source-free alternative. However, challenges arise in mastering cross-modal embeddings and handling user-specific sketch styles. Our proposed framework overcomes these hurdles with a prototypical setup, combined with a grid-based locator and prototypical domain adaptation. We also demonstrate success in few-shot convergence across novel keypoints and classes through extensive experiments.