Grapevine Nutritional Disorder Detection Using Image Processing

Vine nutrition is a key element of vineyard management. Nutrient disorders affect vine growth, crop yield, berry composition, and wine quality. Each vineyard may have a unique combination of soil type, vine age, canopy architecture, cultivar and rootstock. Therefore nutritional requirements vary between vineyards and even locations within a vineyard. Nutritional disorders can be detected visually on leaves, fruits, stems or roots. The advancement of image processing and machine learning has made it feasible to develop rapid tools to assess vine nutritional disorders using these symptoms. This paper presents our proposed method of using a smartphone app to capture and analyse images of vine leaves for identifying nutritional disorders of grapevines rapidly and conveniently. Nutrient deficiency/toxicity symptoms were created in hydroponically grown grapevines of both red and white varieties. RGB (red, green, and blue) images of old and young leaves were taken weekly to track the progression of symptoms. A benchmarked dataset was developed through a laboratory based nutrient analysis of the petioles. A wide range of features (e.g., texture, smoothness, contrast and shape) were selected for the following customised machine learning techniques. Our proposed algorithm was developed to identify specific deficiency and toxicity symptoms through training and testing process. The support vector machine has achieved a 98.99% average accuracy in the testing.