et al., 2017 Khan et al., 2018 Liu et al., 2018 Geetharamani and Pandian, 2019 Ji et al., 2019 Jiang et al., 2019 Liang et al., 2019 Oppenheim et al., 2019 Pu et al., 2019 Ramcharan et al., 2019 Wagh et al., 2019 Zhang et al., 2019a Zhang et al., 2019b ), CNNs are extensively studied and applied to the diagnosis of plant diseases. In ( Mohanty et al., 2016 Zhang and Wang, 2016 Lu J. Thus, using CNNs to identify early plant diseases has become a research focus of agricultural informatization. CNN is still considered to be one of the optimal algorithms for pattern recognition tasks. In contrast, convolutional neural network (CNN) can effectively avoid complex image pre-processing and employ shared weights to reduce memory consumption. However, the classification features in these approaches are selected based on human experience, which limits the generalizability of the models and the accuracies of these models are still not satisfy the recognition requirement. With the development of computer vision technique, researchers have proposed some plant disease recognition algorithms based on machine learning methods ( Waghmare et al., 2016 Ali et al., 2017 Hamuda et al., 2017 Akbarzadeh et al., 2018 Griffel et al., 2018 Sharif et al., 2018 Kaur et al., 2019 Khan et al., 2019 Kour and Arora, 2019 Liu et al., 2019 Wang et al., 2019 Zhu et al., 2019 Mohammadpoor et al., 2020). However, the requirement of bulky sensors and precise instruments leads to low efficiency and high cost ( Mahlein et al., 2013 Lin et al., 2014). Hence, various spectroscopy techniques have been widely applied in plant disease diagnosis and monitoring. The resulting erroneous diagnosis will lead to the abuse of pesticides, which will destroy the growth environment of the grapes and damage the quality of the fruit. However, it not only is visual recognition a time-consuming and laborious task, but the recognition accuracy does not satisfy the requirement ( Dutot et al., 2013). The current approaches for disease detection are based mainly on visual recognition. Hence, the identification and diagnosis of grape leaf diseases have received extensive attention from orchard workers and experts on disease and pest control. However, diseases in grape leaves have hindered the development of the grape industry and caused significant economic losses. The grape industry is one of the major fruit industries in China, and the total output of grapes reached 13.083 million tons in 2017. Meanwhile, this study explores a new approach for the rapid and accurate diagnosis of plant diseases that establishes a theoretical foundation for the application of deep learning in the field of agricultural information. The experimental results demonstrate that the proposed model can efficiently recognize grape leaf diseases. Compared to GoogLeNet and ResNet-34, the recognition accuracy increases by 2.97% and 2.55%, respectively. It realizes an overall accuracy of 97.22% under the hold-out test set. Ultimately, a novel CNN-based model, namely, DICNN, is built and trained from scratch. In addition, a dense connectivity strategy is introduced to encourage feature reuse and strengthen feature propagation. Afterward, Inception structure is applied for strengthening the performance of multi-dimensional feature extraction. First, based on 4,023 images collected in the field and 3,646 images collected from public data sets, a data set of 107,366 grape leaf images is generated via image enhancement techniques. This paper proposes a novel recognition approach that is based on improved convolutional neural networks for the diagnoses of grape leaf diseases. Timely diagnosis and accurate identification of grape leaf diseases are decisive for controlling the spread of disease and ensuring the healthy development of the grape industry. 6West Electronic Business, Co., Ltd., Yinchuan, ChinaĪnthracnose, brown spot, mites, black rot, downy mildew, and leaf blight are six common grape leaf pests and diseases, which cause severe economic losses to the grape industry.5Ningxia Smart Agricultural Industry Technology Collaborative Innovation Center, Yinchuan, China.4College of Mechanical and Electronic Engineering, Northwest A&F University, Yangling, China.3Shaanxi Key Laboratory of Agricultural Information Perception and Intelligent Service, Northwest A&F University, Yangling, China.2Key Laboratory of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Northwest A&F University, Yangling, China.1College of Information Engineering, Northwest A&F University, Yangling, China.Bin Liu 1,2,3*†, Zefeng Ding 1†, Liangliang Tian 1, Dongjian He 2,3,4, Shuqin Li 1,2,5 and Hongyan Wang 5,6
0 Comments
Leave a Reply. |