A research project to develop a portable sensor capable of assessing food quality has won the Buy West Eat Best Kim Chance Fellowship Award from the WA State Government.
University of Western Australia PhD student Jorge Silva was awarded the Fellowship in late 2019 for his research project to develop low-cost spectroscopic infrared scanning technology to assess food quality.
The Kim Chance Fellowship Award was established in recognition of Mr Chance’s devotion to agriculture, and for his tireless support of farmers, the wider agricultural sector and regional communities.
The $10,000 fellowship, funded through the WA Department of Primary Industries and Regional Development, allowed Jorge to travel to the University of Georgia, USA in early 2020 to gain hands on experience in food quality and safety assessment using infrared scanning technology.
Jorge’s innovative ANFF-WA-enabled work that helped to secure the award is aimed at developing a portable spectroscopy system for monitoring the quality of food across the supply chain. The project will improve the quality control measures used to detect faecal contamination in poultry and surface contamination on vegetable crops by looking at two narrow spectral bands centred at 517 nm and 565 nm.
The approach first involves the use of two cameras to calculate the volume of chicken using stereo estimation. Stereo estimation, which is a form of artificial binocular vision, compares the distances of an object from the two cameras to generate a disparity map. This map then allows the system to estimate the volume of the detected object ¬ñ in this case chickens. The result is a contact free approach to classify chicken by size that can happen inline during processing.
At the same time as classifying the chicken, the cameras capture spectral faecal matter signatures using a narrowband optical filter, one tailored to the 517 nm band, while the other is specific to observe 565 nm light. The amplitude of spectral signal allows the system to observe whether faecal matter is present – if there is any, alarm bells start ringing.
The approach has potential to be modified for use in other farming applications. This is possible considering that many spectral features of plants and minerals can be detected within the visible and near infrared bands that can be detected using CMOS cameras commonly found in smartphones and computers.
Preliminary trials will be undertaken with poultry within the next two years, with the aim of having the technology to market within four to five years.