Forearm Gesture Dataset: Gesture Recognition under Different Arm Postures and Force Levels

_{We present a forearm gesture dataset named "}_FEMG-VPF_{", comprising simultaneously recorded high-density sEMG, sparse multi-channel sEMG, and Euler angle signals of the forearm muscles. High-density sEMG signals were recorded with a 64-channel high-density grid as well as sparse 8-channel sEMG and 3-channel Euler angle signals were recorded with a Myo armband. Specifically, 28 healthy subjects performed 38 hand gestures under two distinct forearm states: 1) the upper arm naturally drooping with the forearm parallel to the ground, and 2) the upper arm naturally drooping with the forearm at a 45-degree angle to the ground. Each arm posture incorporates one resting gesture, 15 finger movements, and 3 grasping gestures. The finger movements encompassed in the database include single-finger movement, double-finger movement, triple-finger movement, and full-hand movements. The grasping gestures incorporate actions grasping cylindrical objects (i.e., mineral water bottles), planar objects (i.e., books), and spherical objects (i.e., tennis balls), respectively. Each gesture was performed in 6 repeated trials, namely four high-force trials and two low-force trials.}