IntegratedPIFu: Integrated Pixel Aligned Implicit Function for Single-view Human Reconstruction

TL;DR

IntegratedPIFu is a novel pixel aligned implicit model that leverages depth and human parsing to improve single-view human reconstruction, reducing artifacts and enhancing structural accuracy with fewer parameters.

Contribution

It introduces depth oriented sampling and a new architecture that outperforms existing methods while using fewer model parameters.

Findings

Significantly outperforms state-of-the-art methods.

Reduces noise artifacts in reconstructions.

Improves structural correctness of meshes.

Abstract

We propose IntegratedPIFu, a new pixel aligned implicit model that builds on the foundation set by PIFuHD. IntegratedPIFu shows how depth and human parsing information can be predicted and capitalised upon in a pixel-aligned implicit model. In addition, IntegratedPIFu introduces depth oriented sampling, a novel training scheme that improve any pixel aligned implicit model ability to reconstruct important human features without noisy artefacts. Lastly, IntegratedPIFu presents a new architecture that, despite using less model parameters than PIFuHD, is able to improves the structural correctness of reconstructed meshes. Our results show that IntegratedPIFu significantly outperforms existing state of the arts methods on single view human reconstruction. Our code has been made available online.