Online Adaptive Disparity Estimation for Dynamic Scenes in Structured Light Systems
This work addresses a domain-specific issue for structured light systems in dynamic scenes, offering an incremental improvement in adaptation efficiency.
The paper tackles the problem of performance drop in deep neural networks for dense disparity estimation in unseen environments for monocular structured light systems, proposing an unsupervised loss function based on long sequential inputs that improves online adaptation speed and achieves superior performance on unseen data.
In recent years, deep neural networks have shown remarkable progress in dense disparity estimation from dynamic scenes in monocular structured light systems. However, their performance significantly drops when applied in unseen environments. To address this issue, self-supervised online adaptation has been proposed as a solution to bridge this performance gap. Unlike traditional fine-tuning processes, online adaptation performs test-time optimization to adapt networks to new domains. Therefore, achieving fast convergence during the adaptation process is critical for attaining satisfactory accuracy. In this paper, we propose an unsupervised loss function based on long sequential inputs. It ensures better gradient directions and faster convergence. Our loss function is designed using a multi-frame pattern flow, which comprises a set of sparse trajectories of the projected pattern along the sequence. We estimate the sparse pseudo ground truth with a confidence mask using a filter-based method, which guides the online adaptation process. Our proposed framework significantly improves the online adaptation speed and achieves superior performance on unseen data.