Performance Analysis of Machine Learning Models for Hand Gesture Recognition from Electromyographic Signals

• Lucas Lemos Cerqueira de Freitas UFAL
• Artur Brederodes da Costa Neto UFAL
• Gabriel Lucas Bento Germano UFAL
• Maria Fernanda Herculano Machado da Silva UFMG
• Rodrigo Santos da Silva UFMG
• Thiago Damasceno Cordeiro UFAL

DOI: https://doi.org/10.5753/sbcas.2026.21645

Resumo

Amputation of the upper extremity significantly affects human autonomy by compromising motor functionality, limiting the execution of activities of daily living, and reducing the individual’s capacity for independent interaction with the environment. In 2017, more than 57 million people were living with some form of traumatic amputation, approximately 30% involving the upper limbs. In this context, prostheses controlled by surface electromyographic (sEMG) signals are a promising alternative for restoring motor function, using residual muscles’ electrical activity as a control source. However, automatic gesture recognition from sEMG signals remains challenging, due to physiological variability, noise, and the complexity of neuromuscular patterns. This work presents a systematic and reproducible comparative analysis of four supervised classifiers — KNN, Random Forest, SVM, and Multilayer Perceptron — applied to the NinaPro DB5 dataset, with focus on the joint effect of window size and temporal stride on classification performance, a combination rarely explored in the literature. Time-domain feature extraction and sliding-window segmentation were applied to the models. The KNN and RF models achieved the best results, with accuracies exceeding 99% in certain configurations, reinforcing the potential of supervised methods for developing more accessible and responsive robotic prostheses and human-machine interfaces.

Referências

Atzori, M., Gijsberts, A., Heynen, S., Hager, A.-G. M., Deriaz, O., Van Der Smagt, P., Castellini, C., Caputo, B., and Müller, H. (2012). Building the ninapro database: A resource for the biorobotics community. In 2012 4th IEEE RAS & EMBS International Conference on Biomedical Robotics and Biomechatronics (BioRob), pages 1258–1265. IEEE.

Corbett, E. A., Perreault, E. J., and Kuiken, T. A. (2011). Comparison of electromyography and force as interfaces for prosthetic control. Journal of rehabilitation research and development, 48(6):629–638.

De Luca, C. J. (2002). Surface electromyography: Detection and recording. Technical report, DelSys Incorporated, Boston, USA.

Garcia, M. C. and Vieira, T. (2011). Surface electromyography: Why, when and how to use it. Revista andaluza de medicina del deporte, 4(1):17–28.

Gopal, P., Gesta, A., and Mohebbi, A. (2022). A systematic study on electromyography-based hand gesture recognition for assistive robots using deep learning and machine learning models. Sensors, 22(10):3650.

Kadavath, M. R. K., Nasor, M., and Imran, A. (2024). Enhanced hand gesture recognition with surface electromyogram and machine learning. Sensors, 24(16):5231.

McDonald, C. L., Westcott-McCoy, S., Weaver, M. R., Haagsma, J., and Kartin, D. (2021). Global prevalence of traumatic non-fatal limb amputation. Prosthetics and orthotics international, 45(2):105–114.

Mills, K. R. (2005). The basics of electromyography. Journal of Neurology, Neurosurgery & Psychiatry, 76(suppl 2):ii32–ii35.

Oskoei, M. A. and Hu, H. (2007). Myoelectric control systems—a survey. Biomedical signal processing and control, 2(4):275–294.

Phinyomark, A., Phukpattaranont, P., and Limsakul, C. (2012). Feature reduction and selection for emg signal classification. Expert Systems with Applications, 39(8):7420–7431.

Tanaka, T., Nambu, I., Maruyama, Y., and Wada, Y. (2022). Sliding-window normalization to improve the performance of machine-learning models for real-time motion prediction using electromyography. Sensors, 22(13):5005.