Author(s)

Tianci Liu and Xueyun Chen

Corresponding Author

Tianci Liu

ABSTRACT

At present, the mainstream methods for detecting crop diseases rely mainly on using color and artificially designed feature operators, and the detection accuracy is not high. The state-of-the-art deep learning methods require a large number of training sample images. Relying on the traditional Generative Adversarial Network is difficult to generate images that meet the quality requirements. This paper proposes a light mask Generative Adversarial Network(LMGAN) to identify generate grape leaf diseases images with controllable shape and light intensity. By extracting texture and brightness information of the leaf, the corresponding morphological mask label and light mask label are made and used as the training input of the network. Using the convolutional neural network to detect grape leaves disease. The experimental results show that the proposed algorithm achieves better accuracy than the existing mainstream algorithms.

KEYWORDS

Convolutional neural network; Generative Adversarial Network(GAN); deep learning; crop disease