[S-RD24] Cutting-edge sensing technology applied to geology and resource exploration

[SRD24-P01] Hyperspectral imagery machine learning analyser app for mineral identification: APiS (AI powered intelligence spectrum analyser)

*岡田 夏男¹、滝澤 海斗¹、Okhala Muacanhia¹、若江 史生¹、大友 陽子¹、川村 洋平¹ (1.北海道大学大学院工学院)

キーワード：分光法、ソフトウェア、鉱物、ハイパースペクトルイメージング

We introduce APiS (AI Powered intelligence Spectrum analyzer), an application that integrates hyperspectral imaging with advanced machine learning for mineral processing. APiS provides a user-friendly interface covering data visualization, interactive labeling, model training, and segmentation/testing, all without requiring programming skills. Users can choose from 34 classification and 27 regression models, including 1D-CNN, enabling tasks such as mineral species identification and composition estimation. Experimental results indicate high accuracy, though challenges remain for similar spectra (e.g., chalcopyrite vs. pyrite) and large-scale data processing. Future work will focus on improved data handling and enhanced output features, further expanding APiS’s potential for researchers and industry professionals seeking efficient hyperspectral solutions.