Coffee Berry pathogen anomaly detection using colour and shape separation via L-systems

Coffee berries are susceptible to infection from several sources including fungal diseases, bacterial diseases, and insect pests. The early recognition of these infection sources forms a vital factor in the coffee berry industry, ensuring higher levels of quality and creating the right conditions to support a resilient coffee bean production industry. This paper examines the use of L-systems to allow for the early recognition of pathogens during the various stages of cultivation and processing. This paper introduces a processing method that mimics human vision, using minimum prior knowledge in concert with the separation of colours and the convergence of colour and shape awareness. This process relies upon additional learned knowledge from one or more edge samples that can be extricated from berry images. This system uses coloured lattice squares to discover the size, shape and number of berries as part of the anomaly detection procedure. When used in combination with L-systems plant modelling it demonstrates an effective means to detect the presence of dangerous pathogens such as coffee berry borers (CBBs).