Adversarial learning for viewpoints invariant 3D human pose estimation.

•We propose an adversarial learning framework, which can learn human pose latent from 3D annotated datasets to optimized the estimation of monocular images with only 2D annotations.•We add a viewpoints invariant module to automatically regulate observation viewpoints for generated 3D pose.•We achieved the state-of-art result on Human3.6M dataset and quantitatively improved the 2D pose estimation performance.