Deep-Learning-Based Automated Analysis of Droplet-Impact Splashing Behavior and High-Precision Boundary Estimation
深層学習を用いた液滴衝突スプラッシュ挙動の自動解析と高精度境界線推定の研究
Conventional analysis of droplet splashing — particularly in the small-droplet regime — has relied on visual inspection and manual measurement, limiting throughput, reproducibility, and objectivity. This project develops a deep-learning pipeline that automatically detects, classifies, and quantifies splashing morphologies and estimates splash boundaries with high precision from high-speed imaging. The framework will be benchmarked against the systematic experimental dataset built in the doctoral research, enabling objective, statistically robust comparison across a wide range of droplet temperatures and ambient pressures, and providing a transferable tool for the broader thermal-fluids community.
Conventional analysis of droplet splashing — particularly in the small-droplet regime — has relied on visual inspection and manual measurement, limiting throughput, reproducibility, and objectivity. This project develops a deep-learning pipeline that automatically detects, classifies, and quantifies splashing morphologies and estimates splash boundaries with high precision from high-speed imaging. The framework will be benchmarked against the systematic experimental dataset built in the doctoral research, enabling objective, statistically robust comparison across a wide range of droplet temperatures and ambient pressures, and providing a transferable tool for the broader thermal-fluids community.
