← Back to work

Computer vision & lab automation

LLE Vial Analysis

A hardware-integrated vision system that captures extraction vials, identifies liquid layers and menisci, estimates volumes, and generates laboratory reports.

LLE analysis interface with annotated vial detections

Overview

From problem to working system.

LLE Vial Analysis combines a custom camera and lighting assembly with a trained PyTorch detector and a Flask control interface. The system follows a 48-position plate workflow, measures liquid-liquid extraction layers, flags unsuitable samples, and turns each run into annotated images, Excel reports, and structured data.

Inside the project

  • Custom-trained object detection for vials, liquid layers, menisci, residue, and rag layers
  • Motion-aware image capture integrated with a 48-position laboratory workflow
  • Pixel-to-volume calculations, annotated images, and generated Excel reports
  • Custom 3D-printed camera mount and controlled lighting for repeatable inspection