HCMYO-A Dataset

This dataset contains data for use in handwritten character recognition from sEMG. It contains recordings from 3 healthy subjects perforrming handwriting activity on a touchscreen. The recording device is a Thalmic Labs Myo armband device with eight dry electrodes and sampling frequency of 200 Hz. Each user typed a letter (a-z) or a digit (0-9) on a GUI, then started the capture process by clicking on a button, performed handwriting activity using the index finger or a stylus pen, and finally stopped the capture by clicking on a button. Each user wrote 1,000 times each character class, the number of classes was 36. In total, the dataset contains 108,000 recordings. Each recording comprises two files: one file containing sEMG data and one file containing handwriting trajectory. Each file also contains the stroke number at each instant of the trace. Handwriting trajectory data may be useful for visualization or research on handwriting trajectory reconstruction from EMG.