高效且保细节的单目深度估计模型，适合边缘设备部署。
EfficientDepth: A Fast and Detail-Preserving Monocular Depth Estimation Model
- 融合变压器与轻量卷积解码器，提升细节还原能力。
- 在真实与合成数据上联合训练，精度超越多数现有方法。
- 引入LPIPS损失函数，增强对细微结构的感知能力。
单目深度估计（MDE）在机器人、增强现实和自动驾驶中至关重要。尽管近年进展显著，现有方法仍难以兼顾几何一致性、精细细节、对反光等现实挑战的鲁棒性，以及边缘设备的计算效率。为此，我们提出EfficientDepth，结合变压器架构与轻量卷积解码器，并采用双模密度头以实现高保真深度图估计。模型在标注合成数据、真实图像及伪标签真实图像（由高性能MDE方法生成）上联合训练，采用多阶段优化策略提升训练效率，强调几何一致性和细节保留。此外，引入基于LPIPS的损失函数，鼓励网络生成更丰富的细节。实验表明，EfficientDepth性能媲美或优于现有先进模型，同时显著降低计算资源消耗。
原文摘要 · Abstract (English)
Monocular depth estimation (MDE) plays a pivotal role in various computer vision applications, such as robotics, augmented reality, and autonomous driving. Despite recent advancements, existing methods often fail to meet key requirements for 3D reconstruction and view synthesis, including geometric consistency, fine details, robustness to real-world challenges like reflective surfaces, and efficiency for edge devices. To address these challenges, we introduce a novel MDE system, called EfficientDepth, which combines a transformer architecture with a lightweight convolutional decoder, as well as a bimodal density head that allows the network to estimate detailed depth maps. We train our model on a combination of labeled synthetic and real images, as well as pseudo-labeled real images, generated using a high-performing MDE method. Furthermore, we employ a multi-stage optimization strategy to improve training efficiency and produce models that emphasize geometric consistency and fine detail. Finally, in addition to commonly used objectives, we introduce a loss function based on LPIPS to encourage the network to produce detailed depth maps. Experimental results demonstrate that EfficientDepth achieves performance comparable to or better than existing state-of-the-art models, with significantly reduced computational resources.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。