MSGL-Transformer: A Multi-Scale Global-Local Transformer for Rodent Social Behavior Recognition

2026-04-08

AI Summary

This study addresses the challenges of time-consuming and error-prone manual annotation in rodent social behavior recognition by proposing a lightweight multi-scale global–local Transformer model. The method explicitly captures behavioral dynamics across multiple temporal scales through parallel short-range, mid-range, and global attention branches, and incorporates a Behavior-Aware Modulation (BAM) module to enhance discriminative feature representation. As the first approach to achieve cross-dataset generalization within a unified architecture without task-specific fine-tuning, the model attains 75.4% accuracy (F1 = 0.745) on RatSI and 87.1% accuracy (F1 = 0.8745) on CalMS21, significantly outperforming prevailing methods such as TCN, LSTM, and ST-GCN.

Abstract

Recognition of rodent behavior is important for understanding neural and behavioral mechanisms. Traditional manual scoring is time-consuming and prone to human error. We propose MSGL-Transformer, a Multi-Scale Global-Local Transformer for recognizing rodent social behaviors from pose-based temporal sequences. The model employs a lightweight transformer encoder with multi-scale attention to capture motion dynamics across different temporal scales. The architecture integrates parallel short-range, medium-range, and global attention branches to explicitly capture behavior dynamics at multiple temporal scales. We also introduce a Behavior-Aware Modulation (BAM) block, inspired by SE-Networks, which modulates temporal embeddings to emphasize behavior-relevant features prior to attention. We evaluate on two datasets: RatSI (5 behavior classes, 12D pose inputs) and CalMS21 (4 behavior classes, 28D pose inputs). On RatSI, MSGL-Transformer achieves 75.4% mean accuracy and F1-score of 0.745 across nine cross-validation splits, outperforming TCN, LSTM, and Bi-LSTM. On CalMS21, it achieves 87.1% accuracy and F1-score of 0.8745, a +10.7% improvement over HSTWFormer, and outperforms ST-GCN, MS-G3D, CTR-GCN, and STGAT. The same architecture generalizes across both datasets with only input dimensionality and number of classes adjusted.

Problem

• rodent social behavior recognition
• pose-based temporal sequences
• behavior dynamics
• multi-scale modeling
• automated behavior analysis

Innovation

• Multi-Scale Attention
• Global-Local Transformer
• Behavior-Aware Modulation
• Pose-Based Behavior Recognition
• Temporal Dynamics Modeling