NAPS: Integrating pose estimation and tag-based tracking.

Area of Science:

• Computational ethology
• Animal behavior analysis
• Social network dynamics

Background:

• Quantifying animal behavior has advanced significantly with computational ethology.
• Tracking individuals within social groups remains a challenge, often sacrificing pose detail for identity retention, or vice versa.

Purpose of the Study:

• To develop a hybrid tracking framework, NAPS (ArUco Plus SLEAP), that accurately captures fine-grained behaviors while maintaining individual identity in social groups.
• To enable detailed investigation of social dynamics and individual behavioral variation within a group setting.

Main Methods:

• NAPS integrates deep learning-based pose estimation (SLEAP) with unique markers (ArUco) for robust identity persistence.
• The framework was applied to study the social dynamics of the common eastern bumblebee (*Bombus impatiens*).

Main Results:

• NAPS successfully captures finely resolved behaviors and maintains individual identity over time.
• The framework scales to long-duration, high-frame-rate experiments, facilitating detailed behavioral variation analysis within groups.
• Demonstrated application in analyzing the social dynamics of *Bombus impatiens*.

Conclusions:

• NAPS provides a crucial tool for advancing the study of social group behavior and network dynamics.
• This framework enables the collection of critical data for understanding how individual behavioral variations influence collective dynamics in social species.