Early Discovery of Tomato Foliage Diseases Based on Data Provenance and Pattern Recognition

Abstract

Abstract. This work presents an approach focused in enhancing the quality of tomato crops. We are developing and using low cost computational strategies to support early detection of the late blight. Our approach consorts tomatoes cultivars in an experimental field with inexpensive computer-aided resources based on Web and Android mobile tools in which workers collect scouting data and annotations and take images about the state of the crop, and in image filtering techniques and pattern recognition to detect foliage diseases on tomatoes images. In this study, we use provenance metadata about field observations, images and farmers' annotations as well, to improve the efficiency and accuracy of the patterns recognition algorithms. Our identification method achieved a hit rate of 94.12 %, using a reduced set of digital images of the tomato crops.

Using MLP Neural Networks to Detect Late Blight in Brazilian Tomato Crops

“…The goal of this paper is to present novel computerbased techniques that increase the productivity of tomato crops on small properties in the state of Rio de Janeiro. In this study, we expand our previous works (Vianna & Cruz, 2013a, 2013bNunes et al, 2014, Cruz et al, 2015. The approach aims at supporting small farmers to detect late blight, a foliage disease in tomato crops, by using pattern recognition, more specifically based on Multilayer Perceptron (MLP) neural networks.…”