The project addressed the challenge of recycling packaging in the bottled beverage industry by developing, implementing, and optimizing an AI-based quality analysis system. The core objective was to detect and remove defective bottles directly on the production line, thereby reducing waste, improving sustainability, and maintaining high production efficiency. The project aimed for a universal solution adaptable for broader industry use.
The bottled beverage industry faces the dual challenge of ensuring high production efficiency while reducing waste and improving sustainability. Defective bottles—whether due to misprints, cap issues, or incorrect fill levels—often lead to unnecessary plastic waste, energy use, and resource loss. Finding a way to detect and remove these defects early in production is key to supporting the green and digital transition of the industry.
Together with Birštono Mineraliniai Vandenys (BMV) and BENCO Baltic Engineering Company, we developed and implemented an AI-powered computer vision quality analysis system. The solution integrates directly into production lines and operates in real time, detecting defects across three key stages:
Built with edge AI technology (Intel NUC and Raspberry Pi systems paired with industrial cameras), the system maintains full production speeds of 15,000–22,000 bottles per hour without disrupting workflows.
Embracing global challenges for agrifood excellence
Both partners are scaling their ambitions:
This project shows how interregional innovation can accelerate the twin transition—digital and green—and position European SMEs at the forefront of sustainable, competitive manufacturing.