Feature-Level Analysis of EMG Spectrograms for Hand Gesture Classification

This complementary dataset provides feature-level analyses derived from the Electromyography (EMG) Spectrograms Database for Hand Gesture Classification [1]. It contains a tabular dataset available in CSV format, with 408 rows\u2014one per spectrogram\u2014covering three gestures (open hand, closed hand, pinch). Each row includes global statistical descriptors (Mean, Standard Deviation, Maximum, Minimum, Skewness, Kurtosis, Energy, Contrast) and a 256-bin normalized histogram (Bin_0\u2013Bin_255).This dataset offers a structured representation of the original spectrogram images, enabling feature-based machine learning workflows and direct benchmarking against CNN-based approaches.