🤖 AI Summary
To address insufficient accuracy in unsupervised monocular depth estimation for endoscopic images—caused by domain shift and imbalanced local-global feature modeling—this paper proposes a lightweight, efficient depth estimation method tailored to medical endoscopic scenarios. Our approach introduces three key innovations: (1) a novel low-rank random vector adaptation—a LoRA variant—built upon the Depth Anything foundation model for parameter-efficient domain adaptation; (2) a depthwise separable convolutional residual module to compensate for Transformers’ limited capacity in modeling fine-grained local structures; and (3) integration of intrinsic endoscopic geometric constraints to enhance structural consistency. Evaluated on the SCARED and Hamlyn benchmarks, our method achieves state-of-the-art performance with significantly reduced parameter count. It substantially improves intraoperative 3D spatial awareness, delivering more accurate and safer depth perception for minimally invasive surgery.
📝 Abstract
Depth estimation is a cornerstone of 3D reconstruction and plays a vital role in minimally invasive endoscopic surgeries. However, most current depth estimation networks rely on traditional convolutional neural networks, which are limited in their ability to capture global information. Foundation models offer a promising approach to enhance depth estimation, but those models currently available are primarily trained on natural images, leading to suboptimal performance when applied to endoscopic images. In this work, we introduce a novel fine-tuning strategy for the Depth Anything Model and integrate it with an intrinsic-based unsupervised monocular depth estimation framework. Our approach includes a low-rank adaptation technique based on random vectors, which improves the model's adaptability to different scales. Additionally, we propose a residual block built on depthwise separable convolution to compensate for the transformer's limited ability to capture local features. Our experimental results on the SCARED dataset and Hamlyn dataset show that our method achieves state-of-the-art performance while minimizing the number of trainable parameters. Applying this method in minimally invasive endoscopic surgery can enhance surgeons' spatial awareness, thereby improving the precision and safety of the procedures.