Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

EndoDAC: Efficient Adapting Foundation Model for Self-Supervised Depth Estimation from Any Endoscopic Camera

About

Depth estimation plays a crucial role in various tasks within endoscopic surgery, including navigation, surface reconstruction, and augmented reality visualization. Despite the significant achievements of foundation models in vision tasks, including depth estimation, their direct application to the medical domain often results in suboptimal performance. This highlights the need for efficient adaptation methods to adapt these models to endoscopic depth estimation. We propose Endoscopic Depth Any Camera (EndoDAC) which is an efficient self-supervised depth estimation framework that adapts foundation models to endoscopic scenes. Specifically, we develop the Dynamic Vector-Based Low-Rank Adaptation (DV-LoRA) and employ Convolutional Neck blocks to tailor the foundational model to the surgical domain, utilizing remarkably few trainable parameters. Given that camera information is not always accessible, we also introduce a self-supervised adaptation strategy that estimates camera intrinsics using the pose encoder. Our framework is capable of being trained solely on monocular surgical videos from any camera, ensuring minimal training costs. Experiments demonstrate that our approach obtains superior performance even with fewer training epochs and unaware of the ground truth camera intrinsics. Code is available at https://github.com/BeileiCui/EndoDAC.

Beilei Cui, Mobarakol Islam, Long Bai, An Wang, Hongliang Ren• 2024

Related benchmarks

TaskDatasetResultRank
Monocular Depth EstimationSCARED
Abs Rel0.052
24
Depth EstimationSCARED (test)
Abs Rel0.051
21
Monocular Depth EstimationC3VD (test)
Abs Rel0.153
16
Depth EstimationHamlyn 22 videos
Abs Rel0.275
10
Airway Depth EstimationPhantom dataset
Abs Rel0.27
9
Airway Depth EstimationCollected ex-vivo human lung dataset
Depth Con34.58
9
Monocular Depth EstimationSimCol3D (test)
Abs Rel0.17
8
Monocular Depth EstimationCSD (test)
Abs Rel0.215
8
Ego-motion estimationSimCol3D sequence s1 SyntheticColon I (test)
ATE0.2711
6
Ego-motion estimationSimCol3D SyntheticColon I sequence s2 (test)
ATE0.2937
6
Showing 10 of 17 rows

Other info

Code

Follow for update