In recent years, skeleton-based human action recognition has garnered significant research attention, with proposed recognition or segmentation methods typically validated on large-scale coarse-grained action datasets. However, there remains a lack of research on the recognition of small-scale fine-grained human actions using deep learning methods, which have greater practical significance. To address this gap, we propose a novel approach based on heatmap-based pseudo videos and a unified, general model applicable to all modality datasets. Leveraging anthropometric kinematics as prior information, we extract common human motion features among datasets through an ad hoc pre-trained model. To overcome joint mismatch issues, we partition the human skeleton into five parts, a simple yet effective technique for information sharing. Our approach is evaluated on two datasets, including the public Nursing Activities and our self-built Tai Chi Action dataset. Results from linear evaluation protocol and fine-tuned evaluation demonstrate that our pre-trained model effectively captures common motion features among human actions and achieves steady and precise accuracy across all training settings, while mitigating network overfitting. Notably, our model outperforms state-of-the-art models in recognition accuracy when fusing joint and limb modality features along the channel dimension.

In recent years, human action recognition has received increasing attention as a significant function of human–machine interaction. The human skeleton is one of the most effective representations of human actions because it is highly compact and informative. Many recent skeleton-based action recognition methods are based on graph convolutional networks (GCNs) as they preserve the topology of the human skeleton while extracting features. Although many of these methods give impressive results, there are some limitations in robustness, interoperability, and scalability. Furthermore, most of these methods ignore the underlying information of view direction and rely on the model to learn how to adjust the view from training data. In this work, we propose VW-SC3D, a spatial–temporal model with view weighting for skeleton-based action recognition. In brief, our model uses a sparse 3D CNN to extract spatial features for each frame and uses a transformer encoder to obtain temporal information within the frames. Compared to GCN-based methods, our method performs better in extracting spatial–temporal features and is more adaptive to different types of 3D skeleton data. The sparse 3D CNN makes our model more computationally efficient and more flexible. In addition, a learnable view weighting module enhances the robustness of the proposed model against viewpoint changes. A test on two different types of datasets shows a competitive result with SOTA methods, and the performance is even better in view-changing situations.

This paper introduces ‘SkeletonCLIP++’, an extension of our prior work in human action recognition, emphasizing the use of semantic information beyond traditional label-based methods. The innovation, ‘Weighted Frame Integration’ (WFI), shifts video feature computation from averaging to a weighted frame approach, enabling a more nuanced representation of human movements in line with semantic relevance. Another key development, ‘Contrastive Sample Identification’ (CSI), introduces a novel discriminative task within the model. This task involves identifying the most similar negative sample among positive ones, enhancing the model’s ability to distinguish between closely related actions. Incorporating the ‘BERT Text Encoder Integration’ (BTEI) leverages the pre-trained BERT model as our text encoder to refine the model’s performance. Empirical evaluations on HMDB-51, UCF-101, and NTU RGB+D 60 datasets illustrate positive improvements, especially in smaller datasets. ‘SkeletonCLIP++’ thus offers a refined approach to human action recognition, ensuring semantic integrity and detailed differentiation in video data analysis.