新方法实现多动物的姿态估计、识别和跟踪

瑞士联邦理工学院Alexander Mathis、Mackenzie Weygandt Mathis等研究人员合作实现多动物的姿态估计、识别和跟踪。这一研究成果于2022年4月12日在线发表在国际学术期刊《自然—方法学》上。

研究人员表示，估算多种动物的姿态是一个具有挑战性的计算机视觉问题：频繁的互动会导致遮挡，并使检测到的关键点与正确的个体之间的关联变得复杂，而且有高度相似的动物，它们的互动比典型的多人场景更加密切。

为了应对这一挑战，研究人员在DeepLabCut这个开源姿态估计工具箱的基础上，提供了多动物场景所需的高性能动物组装和跟踪功能。此外，研究人员还整合了预测动物身份的能力，以协助追踪（在闭塞的情况下）。研究人员用四个复杂程度不同的数据集来说明了这个框架的实用性，并将其作为未来算法发展的基准。

附：英文原文

Title: Multi-animal pose estimation, identification and tracking with DeepLabCut

Author: Lauer, Jessy, Zhou, Mu, Ye, Shaokai, Menegas, William, Schneider, Steffen, Nath, Tanmay, Rahman, Mohammed Mostafizur, Di Santo, Valentina, Soberanes, Daniel, Feng, Guoping, Murthy, Venkatesh N., Lauder, George, Dulac, Catherine, Mathis, Mackenzie Weygandt, Mathis, Alexander

Issue&Volume: 2022-04-12

Abstract: Estimating the pose of multiple animals is a challenging computer vision problem: frequent interactions cause occlusions and complicate the association of detected keypoints to the correct individuals, as well as having highly similar looking animals that interact more closely than in typical multi-human scenarios. To take up this challenge, we build on DeepLabCut, an open-source pose estimation toolbox, and provide high-performance animal assembly and tracking—features required for multi-animal scenarios. Furthermore, we integrate the ability to predict an animal’s identity to assist tracking (in case of occlusions). We illustrate the power of this framework with four datasets varying in complexity, which we release to serve as a benchmark for future algorithm development.

DOI: 10.1038/s41592-022-01443-0

Source: https://www.nature.com/articles/s41592-022-01443-0