Phase-Interface Instance Segmentation as a Visual Sensor for Laboratory Process Monitoring

2026-03-11

AI Summary

This study addresses the significant challenges in visually monitoring chemical experiments within transparent glassware, where weak phase boundaries and optical artifacts hinder reliable perception. To tackle this, the experimental process is modeled as the temporal evolution of phase interfaces, leading to the creation of the CTG 2.0 dataset. The authors propose LGA-RCM-YOLO, an extension of YOLOv8-seg that integrates a local-global attention mechanism, a rectangular self-calibration module, and a novel color attribute prediction head. This approach achieves, for the first time, instance segmentation of multiphase interfaces and continuous process monitoring. On CTG 2.0, it attains 84.4% AP@0.5 and 58.43% AP@0.5–0.95 for segmentation, along with 98.71% accuracy in color recognition, effectively enabling automated visual perception for representative laboratory procedures such as separatory funnel extraction and crystallization.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Transparent, Interpretable, Explainable ML
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Reliable visual monitoring of chemical experiments remains challenging in transparent glassware, where weak phase boundaries and optical artifacts degrade conventional segmentation. We formulate laboratory phenomena as the time evolution of phase interfaces and introduce the Chemical Transparent Glasses dataset 2.0 (CTG 2.0), a vessel-aware benchmark with 3,668 images, 23 glassware categories, and five multiphase interface types for phase-interface instance segmentation. Building on YOLO11m-seg, we propose LGA-RCM-YOLO, which combines Local-Global Attention (LGA) for robust semantic representation and a Rectangular Self-Calibration Module (RCM) for boundary refinement of thin, elongated interfaces. On CTG 2.0, the proposed model achieves 84.4% AP@0.5 and 58.43% AP@0.5-0.95, improving over the YOLO11m baseline by 6.42 and 8.75 AP points, respectively, while maintaining near real-time inference (13.67 FPS, RTX 3060). An auxiliary color-attribute head further labels liquid instances as colored or colorless with 98.71% precision and 98.32% recall. Finally, we demonstrate continuous process monitoring in separatory-funnel phase separation and crystallization, showing that phase-interface instance segmentation can serve as a practical visual sensor for laboratory automation.

Problem

• phase-interface instance segmentation
• visual monitoring
• transparent glassware
• laboratory automation
• multiphase interfaces

Innovation

• Phase-Interface Instance Segmentation
• Local-Global Attention
• Rectangular Self-Calibration Module
• Chemical Transparent Glasses Dataset
• Laboratory Process Monitoring