A TinyML Approach to Human Activity Recognition

Abstract Human Activity Recognition has been a favorite topic for the scholars not only because of its wide scale acceptance in industry but areas which may help medical and our normal household works as well. Since to make this technology available last person standing queue it is important that models compiled trained field are just high performing optimized such with incurs least overhead. And thus bringing TinyML into picture specialty optimizing model w.r.t. size model, energy consumption, network bandwidth usage etc. Thus work includes using techniques pruning quantization on pre-proposed analyze changes causes w.r.t accuracy size. Our able infer by both Pruning Quantization human activity recognition we can compress up 10 time without hampering severe diversion model. We have taken three UCI-HAR dataset compare outcomes experiment.