Soil sinkholes significantly influence soil degradation, infrastructure vulnerability, and landscape evolution. However, their irregular shapes, combined with interference from shadows and vegetation, make it challenging to accurately quantify their properties using remotely sensed data. In addition, manual annotation can be laborious and costly. In this study, we introduce a novel self-supervised framework for sinkhole segmentation, termed SinkSAM-Net, which integrates traditional topographic computations of closed depressions with an iterative, geometry-aware, prompt-based Segment Anything Model (SAM). We generate high-quality pseudo-labels through pixel-level refinement of sinkhole boundaries by integrating monocular depth information with random prompts augmentation technique named coordinate-wise bounding box jittering (CWBJ). These pseudo-labels iteratively enhance a lightweight EfficientNetV2-UNet target model, ultimately transferring knowledge to a prompt-free, low-parameter, and fast inference model. Our proposed approach achieves approximately 95\% of the performance obtained through manual supervision by human annotators. The framework's performance was evaluated on a large sinkhole database, covering diverse sinkhole dateset-induced sinkholes using both aerial and high-resolution drone imagery. This paper presents the first self-supervised framework for sinkhole segmentation, demonstrating the robustness of foundational models (such as SAM and Depth Anything V2) when combined with prior topographic and geometry knowledge and an iterative self-learning pipeline. SinkSAM-Net has the potential to be trained effectively on extensive unlabeled RGB sinkholes datasets, achieving comparable performance to a supervised model.
Key Features
Prompt Robustness
We introduce a novel Monte Carlo prompt perturbation method (CWBJ), designed to enhance zero-shot SAM segmentation and reduce common boundary errors associated with topographically derived features.
Annotation-Free Learning
We propose a new pipeline that utilizes pseudo-labels generated by SAM, enabling fully self-supervised training
Depth-Guided Prompting
We replace costly elevation sources with monocular depth maps derived from RGB imagery using MDE.
Scalability and Generalization:
Our lightweight model achieves near-supervised performance across diverse environments and previously unseen geographic regions.
Acknowledgment
Supported by the Ministry of Agriculture Chief Scientist
(Grant 16-17-0005, 2022) and a Negev Scholarship from the Kreitman
School, Ben-Gurion University of the Negev.
Cite Us
@article{RAFAELI20251,
title={SinkSAM-Net: Knowledge-driven self-supervised sinkhole segmentation using topographic priors and Segment Anything Model},
author={Osher Rafaeli and Tal Svoray and Ariel Nahlieli},
journal={ISPRS Journal of Photogrammetry and Remote Sensing},
volume={228},
pages={1--15},
year={2025},
issn={0924-2716},
doi={10.1016/j.isprsjprs.2025.06.035},
url={https://www.sciencedirect.com/science/article/pii/S0924271625002618}
}
