Pyramid Feature Attention Network (PFANet) enhances high-level context features and low-level spatial features for accurate monocular depth prediction.
Efficient monocular depth estimation using flow matching for fast and accurate results.
StableCamH enables unsupervised training of monocular depth networks to learn absolute scale and metric accuracy using object size priors.
直接マッピングを使用した高速な単眼深度推定モデルの提案とその効果的な汎化能力に焦点を当てる。
Utilizing near-field lighting for improved monocular depth estimation in endoscopy videos.
F2Depth proposes a self-supervised indoor monocular depth estimation framework using optical flow consistency and feature map synthesis.
Using ViT embeddings improves monocular depth estimation.
FlowDepth proposes a novel self-supervised multi-frame monocular depth estimation framework that decouples dynamic motion flow, applies depth-cue-aware blurring, and introduces a cost-volume sparse loss to address the mismatch problem, unfairness in photometric errors, and depth uncertainty in low-texture regions.
Marigold, a diffusion model-based method, can effectively leverage the rich visual priors captured in modern generative image models to achieve state-of-the-art performance in zero-shot monocular depth estimation across diverse real-world scenes.
Virtually augmenting the NYU Depth V2 dataset with randomly generated 3D objects improves the performance and generalization of deep neural networks for monocular depth estimation.