Allpress Farms is a Cambridgeshire based agricultural business growing crops for retail UK markets. Leek and onion crops are an important part of the Allpress business and optimisation of the harvest offers a considerable opportunity for improving the overall profitability of the business.
Allpress has a programme underway to develop harvest models which offer the opportunity to predict future yield and improve the efficiency of their operation. The project must overcome several hurdles including a manual data collection process which is further complicated by harvest happening across 10 months of the year and the challenges of visualising crops such as leeks due to their thick canopy. If an accurate harvest time model can be effectively implemented the improvement in efficiency and crop production will have a positive and significant impact on the Allpress business.
The Allpress Team approached NPL to help develop an automated and reproducible imaging system that can improve the collection of crop data in this challenging environment. To improve the data gathering, an NPL imaging system was installed on an existing in-field leek harvester to capture leek dimensions as they passed on a conveyor as part of routine sorting. Using machine vision code and a calibration scheme, both developed by the team at NPL, comparisons were made between the automated and manual data collection for harvested crop.
The automated data compared favourably with the manual recording and has enabled ongoing development to extend the code for use with Allpress and has also initiated further projects using computer vision in different industries
