AI-Driven Anomaly Detection for Pharmaceutical Production

Traditional monitoring methods, including manual sensor adjustments or equipment reconfiguration for different vial types (glass or plastic, ranging from 7 to 100 ml), were slow and prone to errors. Implementing AI in a highly regulated pharmaceutical environment added further complexity, as even minor changes require strict documentation, testing, and approval.

Project Goals

The main goal of the project was to stop tipped vials from reaching the filling machine which caused costly downtime. We needed to build a system that could spot anomalies in real time, alerting machine operators, without constant manual adjustments and interfering with the production line.

Directly adding sensors to the production line was not an option, as any modification to the machine’s construction would require costly revalidation. Thus, we implemented a non-invasive solution using a camera as the sensor.

System components:

• Industrial-grade camera: Basler ace acA2440-20gc
• Low-distortion lens: Fujinon HF12XA-5M 2/3″ 12mm
• Edge computing unit optimized for AI (reServer Industrial J4012)
• Custom made stand and cases, designed for easy disinfection
• Light tower with a buzzer
• Custom computer vision algorithms and deep learning models for vial recognition and tracking

The system captures continuous images of vials on the conveyor belt. A convolutional neural network (CNN) processes each frame to:

• Detect tipped vials in real time
• Distinguish between different vial types (after further training)
• Count vials and downtime
• Trigger audio and visual alerts for operators

This project demonstrates how AI can enhance production lines, even under strict regulatory constraints. Using a camera as a sensor, we developed a non-invasive, flexible system that detects tipped vials in real time and monitors machine downtime. It enabled operators to react earlier to incidents, reducing production line downtime and associated delay costs. Its adaptability provides a robust foundation for future AI-driven improvements in the pharmaceutical industry.