Machine learning & image recognition to reduce post-consumer food waste and identify contamination of food waste by non food items

点击翻译按钮获取中文摘要

The global hospitality industry has two significant environmental and economic challenges which are not currently being met. Firstly, $34bn worth of food waste happens every year from post-consumer food waste ("plate waste"). There is currently no easy way to automatically measure and analyse plate waste so that chefs can make data led interventions to reduce this waste. Secondly, for recycling and disposal purposes, the industry is being required to carefully separate their food waste from other non-food products, and faces heavy fines if cross-contamination occurs. There is no in-kitchen real-time solution to help staff identify when non-food items contaminate food waste.These are both growing opportunities which Winnow Solutions Limited (WSL) is in a unique position to tackle. WSL are market leading providers of cutting edge digital solutions that over 1000 commercial kitchens use to accurately monitor and reduce their food waste. However, these existing solutions are focused on pre-consumer food waste, such as spoilage, preparation and overproduction. WSL's computer vision model is trained on recognising non-complex food waste (e.g. fish and chips), but significant work is required to research if it is possible to adapt the model for complex plate waste (e.g. the mixture of food a customer leaves on their plate post buffet) and to provide real-time notifications of contamination by non-food waste items (e.g. plastic crockery).Awaiting Public Project Summary