EMD-HIST-DF handgesture recognition

This repository contains the datasets and source code used in the above study. Data Raw sEMG signals: 8 subjects (S1–S8), 6 channels, 2000 Hz sampling rate, each CSV file includes 6 channel columns + gesture label (0–15). Processed features: Feature sets extracted after Empirical Mode Decomposition (EMD) and/or histogram‑based methods. Each CSV file contains a feature matrix with the last column as the gesture label. Available feature combinations include: HIST (126 dimensions, without EMD) EMD_HIST (504 dimensions) HIST_mDWT_TD (720) HIST_mDWT (600) HIST_TD (624) mDWT_TD (216) mDWT (96) TD (120) MAV (24) All feature matrices are provided for each subject separately. The data are fully anonymized and comply with ethical requirements. Code GCForest.py: Implementation of the Deep Forest classifier. K_val_main.py: Main script for nested 8‑fold cross‑validation, training, and evaluation. Additional scripts for parameter sensitivity and histogram analysis. requirements.txt: List of Python dependencies. For details on the methodology and experimental setup, please refer to the associated manuscript.