Milk quality control requirement evaluation using a handheld near infrared reflectance spectrophotometer and a bespoke mobile application

•Determining cow milk quality at farm level simplifies and improves the production management for dairies.•A handheld NIR spectrophotometer and an artificial intelligence based mobile application enables real-time estimation of cow milk quality parameters.•Neural networks show admirable performance in order to predict cow milk quality parameters from spectral data. This research introduces a novel approach for real-time analysis of individual cow milk samples in order to get an estimation of required quality control parameters such as lactose, protein, fat, and solids-non-fat (SNF), in order to distinguish their concentrations in conventional cow milk. This will permit the classification of milk samples according to their quality, and help to avoid penalties over quality issues in dairy facilities. To fulfil this goal a newly developed mobile application has been implemented, along with a neural network based model fed with spectral data from a handheld near infrared reflectance (NIR) spectrophotometer. With the combination of this application and a portable NIR sensor, milk quality parameters can be estimated by dairy farms on their own premises. The model was obtained by means of the widely used machine learning framework TensorFlow provided by Google Inc. A total of 903 fresh cow milk samples collected over a 3 year period, were used to train and validate the models. The advantages provided by this mobile application at the milking stage allows us to know in real-time the quality control parameters for each cow milk sample, individually. This offers an immediate management change capability along with enhanced decision making potential at farm level, thus leading to the optimisation of the quality of milk production.