OrchardDepth: Precise Metric Depth Estimation of Orchard Scene from Monocular Camera Images

TL;DR

OrchardDepth introduces a specialized monocular depth estimation method tailored for orchard environments, significantly improving accuracy and providing a new retraining approach to enhance depth prediction quality.

Contribution

The paper presents OrchardDepth, a novel approach for precise metric depth estimation in orchard scenes, and a retraining method that enhances depth accuracy by regularizing dense and sparse depth data.

Findings

RMSE reduced from 1.5337 to 0.6738 in orchard depth estimation

New retraining method improves depth prediction accuracy

Fills gap in monocular depth estimation for orchard environments

Abstract

Monocular depth estimation is a rudimentary task in robotic perception. Recently, with the development of more accurate and robust neural network models and different types of datasets, monocular depth estimation has significantly improved performance and efficiency. However, most of the research in this area focuses on very concentrated domains. In particular, most of the benchmarks in outdoor scenarios belong to urban environments for the improvement of autonomous driving devices, and these benchmarks have a massive disparity with the orchard/vineyard environment, which is hardly helpful for research in the primary industry. Therefore, we propose OrchardDepth, which fills the gap in the estimation of the metric depth of the monocular camera in the orchard/vineyard environment. In addition, we present a new retraining method to improve the training result by monitoring the consistent regularization between dense depth maps and sparse points. Our method improves the RMSE of depth estimation in the orchard environment from 1.5337 to 0.6738, proving our method's validation.