Towards Full-parameter and Parameter-efficient Self-learning For Endoscopic Camera Depth Estimation

TL;DR

This paper introduces a full-parameter, parameter-efficient learning framework for endoscopic depth estimation, significantly improving performance over existing adaptation methods by optimizing multiple subspaces simultaneously.

Contribution

It proposes a novel two-stage framework that adapts attention, convolution, and MLP subspaces and introduces a memory-efficient optimization for better depth estimation.

Findings

Performance improved from 10.2% to 4.1% in key metrics.

Initial experiments show significant accuracy gains.

Framework outperforms state-of-the-art models.

Abstract

Adaptation methods are developed to adapt depth foundation models to endoscopic depth estimation recently. However, such approaches typically under-perform training since they limit the parameter search to a low-rank subspace and alter the training dynamics. Therefore, we propose a full-parameter and parameter-efficient learning framework for endoscopic depth estimation. At the first stage, the subspace of attention, convolution and multi-layer perception are adapted simultaneously within different sub-spaces. At the second stage, a memory-efficient optimization is proposed for subspace composition and the performance is further improved in the united sub-space. Initial experiments on the SCARED dataset demonstrate that results at the first stage improves the performance from 10.2% to 4.1% for Sq Rel, Abs Rel, RMSE and RMSE log in the comparison with the state-of-the-art models.