IT informa builds real-time machine learning system for quality control in flower production
Identifying defective products on a production line depends on speed. When the product is a live cut flower moving through a processing facility, manual inspection cannot keep pace with volume, and the cost of errors reaches the customer.
IT informa’s Ola Mattsson worked as Senior C++ Developer and Application Architect at a technology company developing automated quality control for the horticulture industry. He built a system combining camera control on Raspberry Pi devices, real-time image processing via OpenCV and a TensorFlow-based machine learning environment to classify flower quality as flowers moved through the production flow. The system enabled early identification of defective flowers before they reached the end of the line.
Alongside the core classification system, Ola managed the preparation and handling of training data used to develop and refine the ML models, and built APIs connecting the system to frontend applications and business intelligence tools.
The solution enabled real-time quality control at production speed, reducing reliance on manual inspection and supporting improved product quality and reduced waste.
