A broad range of research areas use behavioral analyses. Despite behavior being extensively used, it is quite uncommon for researchers to analyze long-term continuous behavior, especially in social groups, as it is challenging to accurately track individuals and their behaviors. This is even more pressing in social animals such as marmosets, which show many social close contact behaviors while moving in three dimensions. A study in Communications Biology shows a new video-based system combined with Light Detection and Ranging (Lidar) and deep learning to recognize individual marmosets, tracking multiple individuals in a social context. The software was developed to use facial recognition for accurate identification of the individual animals. Then, using Lidar and video recording, the coordinates of each individual were calculated for trajectory tracking. At the same time, an algorithm tracks and classifies individual animal behaviors, especially social behaviors such as grooming. This proof-of-concept study shows the construction and validation of a functional automatic 3D tracking system adapted to marmosets under free-moving conditions. A tracking system that can be used for longer term behavioral analysis to detect behavioral pattern shifts that are characteristic of diseases.

Original reference: Yurimoto, T. et al. Commun. Biol. 7, 216 (2024)

许多研究领域都使用行为分析。尽管行为被广泛使用，但研究人员分析长期连续行为（尤其是在社会群体中）的情况相当罕见，因为准确跟踪个人及其行为具有挑战性。对于狨猴等社会性动物来说，这一点更为紧迫，因为它们在三个维度移动时表现出许多社会密切接触行为。 《通信生物学》的一项研究展示了一种新的基于视频的系统，结合光探测和测距（激光雷达）和深度学习来识别个体狨猴，在社交环境中跟踪多个个体。该软件的开发目的是使用面部识别来准确识别个体动物。然后，使用激光雷达和视频记录，计算每个人的坐标以进行轨迹跟踪。同时，算法会跟踪和分类个体动物行为，尤其是梳理毛发等社会行为。这项概念验证研究展示了适用于狨猴在自由移动条件下的功能性自动 3D 跟踪系统的构建和验证。一种跟踪系统，可用于长期行为分析，以检测疾病特征的行为模式变化。

原始参考文献： Yurimoto, T. et al.交流。生物。 7 , 216 (2024)