MonoPartNeRF:Human Reconstruction from Monocular Video via Part-Based Neural Radiance Fields

TL;DR

MonoPartNeRF introduces a novel part-based neural radiance field framework for monocular human reconstruction, effectively handling complex poses and occlusions through bidirectional deformation, pose embedding, and attention-based appearance modeling.

Contribution

It presents a new framework combining part-based pose embedding, bidirectional deformation, and attention mechanisms to improve monocular human reconstruction quality.

Findings

Outperforms prior methods in joint alignment and texture fidelity.

Effectively handles complex pose variations and occlusions.

Achieves smoother transitions and better structural continuity.

Abstract

In recent years, Neural Radiance Fields (NeRF) have achieved remarkable progress in dynamic human reconstruction and rendering. Part-based rendering paradigms, guided by human segmentation, allow for flexible parameter allocation based on structural complexity, thereby enhancing representational efficiency. However, existing methods still struggle with complex pose variations, often producing unnatural transitions at part boundaries and failing to reconstruct occluded regions accurately in monocular settings. We propose MonoPartNeRF, a novel framework for monocular dynamic human rendering that ensures smooth transitions and robust occlusion recovery. First, we build a bidirectional deformation model that combines rigid and non-rigid transformations to establish a continuous, reversible mapping between observation and canonical spaces. Sampling points are projected into a parameterized surface-time space (u, v, t) to better capture non-rigid motion. A consistency loss further suppresses deformation-induced artifacts and discontinuities. We introduce a part-based pose embedding mechanism that decomposes global pose vectors into local joint embeddings based on body regions. This is combined with keyframe pose retrieval and interpolation, along three orthogonal directions, to guide pose-aware feature sampling. A learnable appearance code is integrated via attention to model dynamic texture changes effectively. Experiments on the ZJU-MoCap and MonoCap datasets demonstrate that our method significantly outperforms prior approaches under complex pose and occlusion conditions, achieving superior joint alignment, texture fidelity, and structural continuity.