AI Research

Applying artificial intelligence to analytical chemistry and natural-medicine research.

Alongside instrument-based methods, I develop AI tools that turn routine analytical data into fast, reproducible, quantitative results. Each project below is open source.

TLC-RAPID

TLC-RAPID

2026-09

Automated TLC spot detection and quantification from plate images

A deep-learning tool that locates thin-layer chromatography (TLC) spots in digital plate images with a YOLOv5 instance-segmentation model, then back-calculates sample amounts from same-plate standard calibration. It handles visible-light, 254 nm, and 366 nm fluorescence plates, lets users add missed spots by hand, and exports traceable Excel reports. Ships as a one-click Windows application.

99.2%
Precision
95.4%
Recall
97.3%
F1-score

Independent test set of 82 images / 745 bands

Deep learningComputer visionInstance segmentationTLCQuantitative analysis