Domain-Guided Prompting of the Segment Anything Model for Seismic Interpretation: The Role of Attributes, Visualization, and Hybrid Prompts

2026-06-14 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computer Vision & Pattern Recognition, Geophysical AI Applications · Depth: Expert, quick

Summary

A new framework introduces a principled approach for zero-shot adaptation of the Segment Anything Model (SAM) to seismic interpretation, addressing the limitations of fine-tuning which requires extensive labeled data and compromises generalization. This framework integrates two core components: first, aligning seismic attributes and visualization choices, such as colormaps, with specific geological targets; and second, employing a hybrid prompting strategy that combines sparse user-defined point prompts with dense mask prompts derived from SAM's internal feature activations. Systematic evaluation across multiple geological targets, datasets, prompt configurations, and seismic attribute representations demonstrates that this combined approach enhances the separability of geological features, improves boundary delineation, and achieves competitive segmentation accuracy in a fully zero-shot setting. This eliminates the need to retrain SAM for each geological feature, offering a practical and scalable pathway for foundation models in seismic interpretation, reducing reliance on labeled data while preserving model generality.

Key takeaway

For geophysicists and AI scientists interpreting seismic data, this framework offers a robust method to apply foundation models without extensive retraining. You should integrate geological target-aware seismic attributes and colormaps with hybrid prompting to achieve accurate zero-shot segmentation. This approach significantly reduces reliance on labeled data, allowing you to deploy SAM more broadly and efficiently across diverse geological features.

Key insights

Zero-shot SAM adaptation for seismic interpretation is achieved by combining domain-guided visualization with hybrid prompting.

Principles

• Align seismic attributes and colormaps with geological targets.
• Combine sparse point prompts with dense mask prompts.
• Zero-shot adaptation preserves model generality.

Method

The framework involves selecting geological target-aware seismic attributes and colormaps, then applying a hybrid prompting strategy using user-defined points and SAM's internal feature activations for segmentation.

In practice

• Apply hybrid prompting for improved boundary delineation.
• Select specific seismic attributes for feature separability.
• Use colormaps tailored to geological features.

Topics

• Segment Anything Model
• Seismic Interpretation
• Zero-shot Learning
• Foundation Models
• Hybrid Prompting
• Computer Vision

Best for: Computer Vision Engineer, AI Scientist, Research Scientist, Prompt Engineer

Related on AIssential

• From Boundaries to Semantics: Prompt-Guided Multi-Task Learning for Petrographic Thin-section Segmentation — Petro-SAM enables high-quality joint grain-edge and lithology segmentation in petrographic images by adapting SAM.
• DiffuSAM: Diffusion Guided Zero-Shot Object Grounding for Remote Sensing Imagery — DiffuSAM combines diffusion models with segmentation models for superior zero-shot object grounding in remote sensing.
• Metadata-Aware Multi-Prompt Reasoning for Zero-Shot Accident Understanding — Decomposing zero-shot video understanding into distinct stages improves reasoning reliability.
• When Less Is More: Simplicity Beats Complexity for Physics-Constrained InSAR Phase Unwrapping — Simpler convolutional architectures outperform complex attention-based models for physics-constrained geophysical regression tasks.
• MapSR: Prompt-Driven Land Cover Map Super-Resolution via Vision Foundation Models — MapSR uses prompt-driven vision foundation models for efficient, training-free land-cover super-resolution with minimal annotation.
• SAM 3: Concept-Based Visual Understanding and Segmentation — SAM 3 unifies detection, segmentation, and tracking of open-vocabulary concepts using text or visual prompts.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.