The Segment Anything Model (SAM), a large pretrained foundation model for computer vision, offers powerful zero-shot segmentation capabilities through prompt-based interaction, making it a promising tool for seismic interpretation. However, most existing applications rely on fine-tuning for specific geological targets, which requires extensive labeled data, incurs high computational cost, and often compromises the model's generalization capability, according to a study published on arXiv.
To overcome these limitations, the authors introduce a principled framework for zero-shot adaptation of foundation models to seismic data. The framework is built on two key components: (1) aligning seismic attributes and visualization choices (e.g., colormaps) with the geological target of interest, and (2) employing a hybrid prompting strategy that combines sparse user-defined point prompts with dense mask prompts derived from SAM's internal feature activations.
The study systematically evaluates this framework across multiple geological targets, datasets, prompt configurations, and seismic attribute representations. Results demonstrate that geologic target-aware selection of seismic attributes and colormaps, combined with hybrid prompting, enhances the separability of geological features and improves boundary delineation and segmentation accuracy relative to point-based prompting alone.
| Component | Description | Role in Framework |
|---|---|---|
| Seismic attributes & colormaps | Aligned with specific geological targets | Enhances feature separability and boundary delineation |
| Hybrid prompting | Combines sparse user-defined points with dense mask prompts from SAM internal activations | Improves segmentation accuracy over point-only prompts |
| Zero-shot setting | No retraining of SAM | Reduces dependency on labeled data and computational cost |
A key finding is that when these components are jointly applied, SAM can achieve competitive segmentation performance in a fully zero-shot setting, thereby eliminating the need to retrain SAM for each geologic feature, the authors report. This establishes a practical and scalable pathway to leverage foundation models in seismic interpretation, reducing reliance on labeled data while preserving model generality.
For enterprise technology decision-makers, this work illustrates a broader principle: domain-specific cues (attributes, visualizations) and hybrid prompting can unlock the power of large foundation models without the prohibitive costs of fine-tuning. Although the current focus is geoscience, the same approach could be adapted to other fields where labeled data is scarce and models must generalize across diverse targets.
