Digging into contrastive learning for robust depth estimation with diffusion models

arXiv:2404.09831v1 Announce Type: new
Abstract: Recently, diffusion-based depth estimation methods have drawn widespread attention due to their elegant denoising patterns and promising performance. However, they are typically unreliable under adverse conditions prevalent in real-world scenarios, such as rainy, snowy, etc. In this paper, we propose a novel robust depth estimation method called D4RD, featuring a custom contrastive learning mode tailored for diffusion models to mitigate performance degradation in complex environments. Concretely, we integrate the strength of knowledge distillation into contrastive …

abstract arxiv attention cs.cv denoising diffusion diffusion models etc however novel paper patterns performance robust type world