KPM subsdidiary Smart Vision Works has launched SiftAI, an AI-based vision inspection system to improve foreign object detection and quality grading in food processing. The new system is designed to detect and remove foreign objects at greater speed than current approaches.

The SiftAI Smart Table is trained to automatically detect and remove foreign objects and sort products in a single pass. Beta customers have reported higher produce profit, fewer missed contaminants, and dramatically lower labour costs.

The system includes cameras, AI software, a conveyor, and automatic ejection mechanisms with dual drops (one for foreign material, one for culls) to ensure only ideal potatoes reach later process stages.

The system can help processors avoid customer fines for contaminants. For some processors, these fines in the US may amount to $50,000 per incident and total hundreds of thousands of dollars per year. 

Because of the speed of the AI technology, conveyor speeds can be increased, giving operators more throughput per shift. 

The system precisely grades produce using sophisticated evaluation of bruises, rot, and green. According to Christopher Bryant, President of the Smart Vision Works AI Division within KPM Analytics, “we estimate that up to 20% of potatoes are typically diverted to the wrong value stream, reducing profitability and causing customer satisfaction issues.” A more accurate sorting system diverts less produce to less profitable uses, and higher quality produce can command a higher price.