The field of gait analysis and clustering is moving towards more scalable and generalizable models. Recent developments have focused on leveraging large language models (LLMs) and multimodal fusion techniques to improve the accuracy and interpretability of gait recognition systems. Notably, researchers are exploring the use of LLMs to incorporate domain knowledge and user preferences into clustering algorithms, enabling more flexible and adaptable models. Additionally, the integration of multimodal data, such as RGB and Depth (RGB-D) inputs, has shown promising results in improving the robustness and accuracy of gait analysis systems.

Some particularly noteworthy papers in this area include: FoundationGait, which introduces a scalable, self-supervised pretraining framework for gait understanding, achieving state-of-the-art performance on several benchmarks. Explainable Parkinsons Disease Gait Recognition Using Multimodal RGB-D Fusion and Large Language Models, which presents an explainable multimodal framework for recognizing Parkinsonian gait patterns, demonstrating higher recognition accuracy and improved robustness to environmental variations. ClusterFusion, which proposes a hybrid clustering framework that leverages LLMs as the clustering core, guided by lightweight embedding methods, achieving state-of-the-art performance on standard tasks and substantial gains in specialized domains.