Winnow Vision could take photos when the food throws in the bin. In the training phase, the system shortlists the top 5 predict result of the food, then the users required to confirm the right result. With more images are collected, the system could full automatically recognizes the wasted food items. The Winnow Vision could save staff time to identify wasted food and reduce human error. According to the project website, the system users cut their food costs by between 2 and 8% per year.