Hand Gesture Recognition via sEMG signals with CNNs (Electrical and Computer Engineering - MSc Thesis)
The source code for the real-time hand gesture recognition algorithm based on Temporal Muscle Activation maps of multi-channel surface electromyography (sEMG) signals (ICASSP 2021)
Accompaniment code for 'Hilbert sEMG data scanning for hand gesture recognition based on Deep Learning' published in NCAA.
Source code for multiple parameter modelling of synthetic electromyography data.
Biomedical signal (EEG/sEMG/ECG) completion/imputation using diffusion model. "A robust denoising diffusion framework for completing missing regions of multiple biomedical signals"
Computationally-free personalization at test time for sEMG gesture classification. Fast (gpu/cpu) ninapro API.
Auto-learning search framework based on a weighted double Q-learning algorithm:"Integrated block-wise neural network with auto-learning search framework for finger gesture recognition using sEMG signals"
Python algorithm to assess muscle activation patterns during cyclical movements
Cross-subject sEMG lower-limb movement classification on SIAT-LLMD (n=40): per-subject normalization beats learned domain adaptation. 85.8% LOSO macro-F1, externally replicated on ENABL3S.
Robust sEMG-based hand gesture recognition using deep learning and multi-feature signal processing.
Surface-EMG hand-gesture classification for prosthetic control — a clean, leak-free classical-ML pipeline (NinaPro DB2) with a live demo.
Silent Speech Recognition using sEMG and Deep Learning
A low-cost 6-channel sEMG acquisition, gesture recognition, and tendon-driven robotic hand research prototype.
Minimum sEMG electrode configuration for grip force estimation during assisted grasping with a soft robotic glove, using a physics-informed neural network on Ninapro DB2.
Uncertainty-aware single-channel sEMG hand gesture classification with MC Dropout selective prediction — LOSOCV benchmark on YTU dataset and NinaPro DB2 external validation
PyTorch design-space study: 5 sequence models across 3 prediction horizons for sEMG-based variable-impedance teleoperation latency compensation (synthetic data).
Bachelor Thesis work developed in 2025 at University of Bologna. See README for more infos about the project.
Evaluation protocols matter: quantifying protocol sensitivity in sEMG CNNs
Classical machine learning compared on six-gesture surface electromyography, up to leave-one-subject-out.
To associate your repository with the semg topic, visit your repo's landing page and select "manage topics."