4SkeletonFeat: A Framework for Unifying the Extraction of Features from Videos of Human Activities
Resumo
Neste artigo, é proposto um framework para unificar a extração de características de vídeos de atividades humanas. O framework, chamado 4SkeletonFeat, fornece um conjunto de ferramentas e bibliotecas que permitem aos pesquisadores extrair facilmente características de vídeos de atividades humanas. Ele permite extrair características como: ”Joint”, ”Bone”, ”Bone-Motion” e ”Joint-Motion” de vídeos de atividades humanas. Este framework também fornece um conjunto de ferramentas para representação dessas características, como representações baseadas em tokens, representações baseadas em grafos, representações pseudo-imagem e representações baseadas em sequência. Para avaliar o framework proposto, um novo conjunto de dados chamado FitTube foi desenvolvido a partir de vídeos do Youtube. Os resultados mostram que 4SkeletonFeat é capaz de extrair características de vídeos de atividades humanas e fornecer uma boa representação para essas características para tarefas de Aprendizado de Máquina.
Referências
Abdelkawy, A., Ali, A., and Farag, A. A. (2024). Epam-net: An efficient pose-driven attention-guided multimodal network for video action recognition. Neurocomputing, 633:129781.
Alomar, K., Aysel, H. I., and Cai, X. (2024). Cnns, rnns and transformers in human action recognition: a survey and a hybrid model. Artificial Intelligence Review, 58.
Chen, D., Chen, M., Wu, P., Wu, M., Zhang, T., and Li, C. (2025). Two-stream spatio-temporal gcn-transformer networks for skeleton-based action recognition. Scientific Reports, 15.
Cui, H., Huang, R., Zhang, R., and Hayama, T. (2025). Dstsa-gcn: Advancing skeleton-based gesture recognition with semantic-aware spatio-temporal topology modeling. Neurocomputing, 637:130066.
Gao, Z., Chen, J., Liu, Y., Jin, Y., and Tian, D. (2025). A systematic survey on human pose estimation: upstream and downstream tasks, approaches, lightweight models, and prospects. Artificial Intelligence Review, 58.
Halmich, C., Höschler, L., Schranz, C., and Borgelt, C. (2025). Data augmentation of time-series data in human movement biomechanics: A scoping review. PLOS One, 20.
Hu, H., Fang, Y., Han, M., and Qi, X. (2024). Multi-scale adaptive graph convolution network for skeleton-based action recognition. IEEE Access, 12:16868–16880.
Lan, G., Wu, Y., and Hao, Q. (2024). Dir-bhrnet: A lightweight network for real-time vision-based multiperson pose estimation on smartphones. IEEE Transactions on Industrial Informatics, 20:12533–12541.
Liu, M., Liu, H., Hu, Q., Ren, B., Yuan, J., Lin, J., and Wen, J. (2025). 3d skeleton-based action recognition: A review.
Neupane, R. B., Li, K., and Boka, T. F. (2024). A survey on deep 3d human pose estimation. Artificial Intelligence Review, 58.
Pan, L., Lu, J., and Tang, X. (2024). Spatial-temporal graph neural ode networks for skeleton-based action recognition. Scientific Reports, 14.
Pellano, K. N., Strümke, I., and Ihlen, E. A. F. (2024). From movements to metrics: Evaluating explainable ai methods in skeleton-based human activity recognition. Sensors (Basel, Switzerland), 24.
Rode, D., Dunkel, A., Willi, R., Wolf, P., Xiloyannis, M., and Riener, R. (2025). Assessment of monocular human pose estimation models for clinical movement analysis. Scientific Reports, 15.
Shaikh, M. B., Chai, D., Islam, S. M. S., and Akhtar, N. (2024). From cnns to transformers in multimodal human action recognition: A survey. ACM Transactions on Multimedia Computing, Communications and Applications, 20:1 – 24.
Tu, Z., Zhang, Z., Gong, J., Yuan, J., and Du, B. (2025). Informative sample selection model for skeleton-based action recognition with limited training samples. IEEE Transactions on Image Processing, 34:7335–7346.
Wang, H., Weng, W., Wang, J., Zhao, F., Xie, G.-S., Geng, X., and Wang, L. (2026). Foundation model for skeleton-based human action understanding. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(1):47–61.
Xiang, L. and Wang, Z. (2024). Joint mixing data augmentation for skeleton-based action recognition. ACM Transactions on Multimedia Computing, Communications and Applications, 21:1 – 24.
Xie, D., Zhang, X., Gao, X., Zhao, H., and Du, D. (2025). Maf-net: A multimodal data fusion approach for human action recognition. PLOS One, 20.
Yan, J., Zhang, X., Tan, C., and Li, D. (2026). Skelformer: An adaptive hierarchical transformer-based approach on skeleton graphs for human action recognition in video sequences. PLOS ONE, 21(1):1–20.
Zhang, J., Lin, L., Yang, S., and Liu, J. (2024). Self-supervised skeleton-based action representation learning: A benchmark and beyond. International Journal of Computer Vision, 134.
Alomar, K., Aysel, H. I., and Cai, X. (2024). Cnns, rnns and transformers in human action recognition: a survey and a hybrid model. Artificial Intelligence Review, 58.
Chen, D., Chen, M., Wu, P., Wu, M., Zhang, T., and Li, C. (2025). Two-stream spatio-temporal gcn-transformer networks for skeleton-based action recognition. Scientific Reports, 15.
Cui, H., Huang, R., Zhang, R., and Hayama, T. (2025). Dstsa-gcn: Advancing skeleton-based gesture recognition with semantic-aware spatio-temporal topology modeling. Neurocomputing, 637:130066.
Gao, Z., Chen, J., Liu, Y., Jin, Y., and Tian, D. (2025). A systematic survey on human pose estimation: upstream and downstream tasks, approaches, lightweight models, and prospects. Artificial Intelligence Review, 58.
Halmich, C., Höschler, L., Schranz, C., and Borgelt, C. (2025). Data augmentation of time-series data in human movement biomechanics: A scoping review. PLOS One, 20.
Hu, H., Fang, Y., Han, M., and Qi, X. (2024). Multi-scale adaptive graph convolution network for skeleton-based action recognition. IEEE Access, 12:16868–16880.
Lan, G., Wu, Y., and Hao, Q. (2024). Dir-bhrnet: A lightweight network for real-time vision-based multiperson pose estimation on smartphones. IEEE Transactions on Industrial Informatics, 20:12533–12541.
Liu, M., Liu, H., Hu, Q., Ren, B., Yuan, J., Lin, J., and Wen, J. (2025). 3d skeleton-based action recognition: A review.
Neupane, R. B., Li, K., and Boka, T. F. (2024). A survey on deep 3d human pose estimation. Artificial Intelligence Review, 58.
Pan, L., Lu, J., and Tang, X. (2024). Spatial-temporal graph neural ode networks for skeleton-based action recognition. Scientific Reports, 14.
Pellano, K. N., Strümke, I., and Ihlen, E. A. F. (2024). From movements to metrics: Evaluating explainable ai methods in skeleton-based human activity recognition. Sensors (Basel, Switzerland), 24.
Rode, D., Dunkel, A., Willi, R., Wolf, P., Xiloyannis, M., and Riener, R. (2025). Assessment of monocular human pose estimation models for clinical movement analysis. Scientific Reports, 15.
Shaikh, M. B., Chai, D., Islam, S. M. S., and Akhtar, N. (2024). From cnns to transformers in multimodal human action recognition: A survey. ACM Transactions on Multimedia Computing, Communications and Applications, 20:1 – 24.
Tu, Z., Zhang, Z., Gong, J., Yuan, J., and Du, B. (2025). Informative sample selection model for skeleton-based action recognition with limited training samples. IEEE Transactions on Image Processing, 34:7335–7346.
Wang, H., Weng, W., Wang, J., Zhao, F., Xie, G.-S., Geng, X., and Wang, L. (2026). Foundation model for skeleton-based human action understanding. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(1):47–61.
Xiang, L. and Wang, Z. (2024). Joint mixing data augmentation for skeleton-based action recognition. ACM Transactions on Multimedia Computing, Communications and Applications, 21:1 – 24.
Xie, D., Zhang, X., Gao, X., Zhao, H., and Du, D. (2025). Maf-net: A multimodal data fusion approach for human action recognition. PLOS One, 20.
Yan, J., Zhang, X., Tan, C., and Li, D. (2026). Skelformer: An adaptive hierarchical transformer-based approach on skeleton graphs for human action recognition in video sequences. PLOS ONE, 21(1):1–20.
Zhang, J., Lin, L., Yang, S., and Liu, J. (2024). Self-supervised skeleton-based action representation learning: A benchmark and beyond. International Journal of Computer Vision, 134.
Publicado
19/10/2026
Como Citar
GODOI, Diogo H.; ROMERO, Roseli A. F.; RANIERI, Caetano Mazzoni; VARGAS, Patricia A..
4SkeletonFeat: A Framework for Unifying the Extraction of Features from Videos of Human Activities. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 301-312.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.29932.
