WeproposeamethodforhumanposeestimationbasedonDeepNeuralNetworks(DNNs).TheposeestimationisformulatedasaDNN-basedregressionproblemtowardsbodyjoints.WepresentacascadeofsuchDNNregressorswhichresultsinhighprecisionposeestimates.Theapproachhastheadvantageofreasoningaboutposeinaholistic