A high quality dataset for short-duration actions
StrokeRehab is a large-scale, multimodal dataset that serves as a new benchmark for recognizing elemental short-duration actions at high temporal resolution. Designed to address the challenges of automatic action identification from video and kinematic data, StrokeRehab is particularly tailored for applications requiring detailed motion analysis, such as arm rehabilitation post-stroke.
The actions in the StrokeRehab dataset are much shorter (sub-second). They correspond to functional primitives that execute one goal, and are therefore qualitatively different to the higher-level actions in existing benchmark datasets.
StrokeRehab consists of high-quality inertial measurement unit sensor and video data of 51 stroke-impaired patients and 20 healthy subjects performing activities of daily living like feeding, brushing teeth, etc.
Because it contains data from both healthy and impaired individuals, StrokeRehab can be used to study the influence of distribution shift in action-recognition tasks.
In the rehabilitation of arm impairment after stroke, quantifying the training dose (number of repetitions) requires differentiating motions with sub-second durations. StrokeRehab dataset helps to build deep learning models that can different motions with sub-second durations.