Algorithm based on one monocular video delivers highly valid and reliable gait parameters

Abstract Despite its paramount importance for manifold use cases (e.g., in the health care industry, sports, rehabilitation and fitness assessment), sufficiently valid reliable gait parameter measurement is still limited to high-tech laboratories mostly. Here, we demonstrate excellent validity test–retest repeatability of a novel assessment system which built upon modern convolutional neural networks extract three-dimensional skeleton joints from monocular frontal-view videos walking humans. The study based on comparison GAITRite pressure-sensitive walkway system. All measured parameters (gait speed, cadence, step length time) showed concurrent multiple walk trials at normal fast speeds. test–retest-repeatability same level as In conclusion, are convinced that our results can pave way cost, space operationally effective analysis broad mainstream applications. Most sensor-based systems costly, must be operated by extensively trained personnel motion capture systems) or—even if not quite costly—still possess considerable complexity wearable sensors). contrast, video sufficient method presented here obtained anyone, without much training, via smartphone camera.