Recently, 3d face reconstruction has vastly benefited from the surging waves of monocular image-based deep learning methods. however, these methods are still unreliable for effective deployments due to their sensitivity towards the facial occlusions and lack of ability to maintain the identity consistency across the distinct occlusions over the same facial image. to overcome the occlusion sensitivity, we propose distillation assisted mono image occlusion robustification (damior) framework, which leverages the knowledge from occlusion frail trainer (oft) network for attaining the robustness against the facial occlusions. besides, to tackle the issue of identity inconsistency, we propose duplicate images assisted multi occlusions robustification (diamor) framework, which utilizes estimates from damior to subdue the inconsistencies in geometry and texture (collectively known as identity) of the reconstructed 3d faces. the performance of damior is evaluated on the two variations of celeba test dataset: empirical occlusions and irrational occlusions. moreover, we exploit the latter variation to analyze the performance of the proposed diamor framework. our methods outperform state-of-the-art methods by a large margin (for example, diamor reduces 3d vertex-based shape error from 0.670 to 0.394 and texture error from 0.239 to 0.187 for empirical occlusions whereas the shape-error and texture-error for diamor decrease from 0.710 to 0.408 and from 0.285 to 0.198, respectively on the irrationally occluded facial data) demonstrating the efficacy of the proposed methods.