Search

Crop diseases can cause major yield losses, so the ability to detect and identify them in their early stages is important for disease control. Deep learning methods have shown promise in classifying multiple diseases; however, many studies do not use datasets that represent real field conditions, necessitating either further image processing or reducing their applicability. In this paper, we present a dataset of wheat images taken in real growth situations, including both field and glasshouse conditions, with five categories: healthy plants and four foliar diseases, yellow rust, brown rust, powdery mildew and Septoria leaf blotch. This dataset was used to train a deep learning model. The resulting model, named CerealConv, reached a 97.05% classification accuracy. When tested against trained pathologists on a subset of images from the larger dataset, the model delivered an accuracy score 2% higher than the best‐performing pathologist. Image masks were used to show that the model was using the correct information to drive its classifications. These results show that deep learning networks are a viable tool for disease detection and classification in the field, and disease quantification is a logical next step. [ABSTRACT FROM AUTHOR]

Septoria tritici blotch (STB) is one of the main causes of wheat yield loss in the world. Apart from good agrotechnical practice, disease‐resistant cultivars are required to prevent yield losses. Breeding of such cultivars is time‐consuming and laborious, mainly due to the quantitative character of such resistance, but can be shortened using precise phenotyping. We show that fluorescence parameters, collected in a high‐throughput system, can be used for estimating wheat STB resistance. Using machine learning methods (deep learning network, random forest), we demonstrate that disease resistance based on the percentage of necrotic leaf area can be estimated by alterations in fluorescence parameters 8 days after inoculation, and before disease symptoms become visible. Moreover, we applied a random forest classifier to assess the significance of fluorescence parameters in an informative way (classification accuracy = 64.84%, p value [Accuracy >No information rate] <0.001). Based on this, we observed extensive alterations in parameters of sensitive cultivars, reflecting worse photosynthetic performance. The most highly altered parameters were fluorescence intensity measurements and descriptors of energy flux through photosystem II (PSII) reaction centres towards PSI such as F2, F3 and F4 fluorescence intensities at 0.1, 0.270 and 2 ms; initial to maximal fluorescence ratio (F0/Fm); the time derivative of relative variable fluorescence (dV/dT0); maximum quantum yield of PSII in the dark‐adapted state (Fv/Fm); maximum electron transport flux at PSII (ET0/RC); maximum trapped excitonic energy flux per excited PSII cross section (CS) at time T0 (TR0/CS0); and quantum efficiency of the reduction of end acceptors dR0 [dR0/(1 − dR0)]. [ABSTRACT FROM AUTHOR]