Segmenting the Complex and Irregular in Two-Phase Flows: A Real-World Empirical Study with SAM2

2025-08-07

AI Summary

Segmenting non-spherical, deformable, coalescing, or fragmenting bubbles in multiphase flows remains challenging—especially in industrial applications like air lubrication—due to highly non-convex and topologically complex bubble patches that cause failure of conventional methods. Method: This work pioneers the adaptation of the vision foundation model SAM v2.1 to bubble segmentation. Leveraging only 100 annotated images, we employ transfer learning and fine-tuning to enhance modeling of irregular bubble structures in real-world industrial imagery. Contribution/Results: Our approach achieves significantly higher segmentation accuracy than both classical image-processing techniques and state-of-the-art deep learning methods, while maintaining extremely low annotation cost. It demonstrates strong cross-scene generalizability and robustness under varying flow conditions. This establishes a scalable, highly robust paradigm for in-situ perception of complex multiphase flows—enabling practical deployment in demanding industrial environments.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Security and Privacy: Data transparency and provenance

Abstract

Segmenting gas bubbles in multiphase flows is a critical yet unsolved challenge in numerous industrial settings, from metallurgical processing to maritime drag reduction. Traditional approaches-and most recent learning-based methods-assume near-spherical shapes, limiting their effectiveness in regimes where bubbles undergo deformation, coalescence, or breakup. This complexity is particularly evident in air lubrication systems, where coalesced bubbles form amorphous and topologically diverse patches. In this work, we revisit the problem through the lens of modern vision foundation models. We cast the task as a transfer learning problem and demonstrate, for the first time, that a fine-tuned Segment Anything Model SAM v2.1 can accurately segment highly non-convex, irregular bubble structures using as few as 100 annotated images.

Problem

• Segmenting irregular gas bubbles in multiphase flows
• Overcoming limitations of traditional near-spherical shape assumptions
• Applying fine-tuned SAM2 for non-convex bubble segmentation

Innovation

• Uses Segment Anything Model SAM v2.1
• Fine-tuned with 100 annotated images
• Segments irregular bubble structures accurately