Rice is one of the most crucial crops worldwide in modern agriculture. The current diseases (such as rice blasts) have seriously threatened the superior quality and yield of the rice in recent years. However, manual disease monitoring cannot fully meet the large-scale rice cultivation in the extensive fields. Furthermore, the mechanical equipment for disease detection can frequently cause unnecessary harm to the rice plants. As a result, it is highly required for the intelligent and large-scale detection of diseases to reduce the labor costs for the high precision of monitoring. Deep learning models (such as GoogLeNet) can be expected to accurately identify the rice blast. Nevertheless, an optimal balance between detection accuracy and processing speed is very necessary for GoogLeNet in the practical applications of rice blast detection. This study aims to classify and detect the rice blast diseases at the seedling stage using an improved GoogLeNet model. Particularly, the rice was vulnerable to diseases during the tillering stage. Specifically, the research targets were also collected from the rice images in the critical period. The disease features were determined to significantly influence the subsequent rice growth using refined models. Several key steps were also involved: Initially, a comprehensive and diverse dataset of rice blast images was obtained after field research, expert consultations, and advanced techniques of image capture, followed by image processing. Subsequently, data augmentation (including rotation and brightness adjustments) was employed to obtain the final dataset with 2000 rice blast images. An attention mechanism was then integrated into the GoogLeNet model. The distinct features of rice blasts were focused on after optimization. An ablation test was conducted to assess the effectiveness of the improved model. Comparative experiments were also performed to illustrate its advantages. The results indicated that this modification significantly improved the detection accuracy. Specifically, the attention mechanism module (GoogLeNet+DSCAM) was incorporated to increase the recall by 9.29 percentage points, compared with the original model. The attention mechanism has effectively enhanced the detection of critical information. Furthermore, the improved GoogLeNet model also surpassed the original model, in terms of all evaluation metrics. The better performance was achieved to improve by 15.33 percentage points in the precision, with significant gains also observed in recall and F1 score, Thereby the refined network architecture substantially enhanced the detection performance. The superiority of the improved GoogLeNet was observed in the tasks of rice blast classification. Several widely-recognized classification models (including AlexNet, ResNet, VGG, and the original GoogLeNet) were selected as the benchmarks for comparison. The rice blast datasets and an independent test set were utilized to fully train and then evaluate these models. Meanwhile, the comparative analysis showed that there were distinct advantages and practical efficacy of the improved GoogLeNet model. Such classification also demonstrated some challenges. The enhanced GoogLeNet model exhibited exceptional performance overall evaluation metrics, indicating a marked superiority over AlexNet, ResNet, VGG, and the original GoogLeNet. Specifically, the notable improvements were also achieved by 16.11 percentage-point improvement in precision compared to AlexNet and a 15.33 percentage-point improvement compared to the original GoogLeNet, with significant enhancements also observed in recall and F1 score. These significant performances can provide high efficacy to enhance the accuracy and the efficient speed of the detection. This finding can also offer valuable insights into preventing and controlling rice diseases.

Here a dynamic model was developed on the tiller number of the rice population. The double Logistic model was also used to quantitatively analyze the dynamic process of the tiller occurrence and the extinction. A set of indicators was defined to describe the tillering dynamics, including the total number of growing tillers(N_g), the total number of dead tillers(N_d), the number of retained tillers(N_r), the start time of tillering(T_st), the peak time of tillering(T_pt), the end time of tillering(T_et), the start time of tillers death(T_sd), the peak time of tillers death(T_pd), the end time of tillers death(T_ed), the duration of tillering(D_t), the duration of tillers death(D_d), the inherent rate of tillering(R_it), the inherent rate of tillers death(R_id), the maximum tillering rate(R_mt), the maximum tillers death rate(R_md). According to the temporal characteristics of the rice tillering, the formula was derived to calculate the indicators of the tillering dynamics. The goodness and adaptability of the model were tested with the dynamic datasets of the rice tillers under different genotypes, transplanting, sowing time, and transplanting density. The model and the indicators were used to explore the dynamic tillering response to the cultivation density. The results were as follows. (1) The model shared the better fitting for the dynamic dataset of rice tillers under different genotypes, transplanting, planting time, and transplanting density. The standard root mean square error (S_RMSE) was followed by the Gamma distribution with the mean was less than 5% and 99% S_RMSE less than 10%. (2) The dynamic indicators and model parameters of tillers after calculation showed a better response to the cultivation density. Taking the planting density test of Huiliangyou 898 as an example, the number of tillers per unit area was accelerated and then slowed down after transplanting at 15 to 33.75 hills/m^2. The number of tillers per unit area reached the peak at 45-50 d after transplanting. After the peak, the number of tillers per unit area decreased rapidly and then slowed down slowly. There was no change in the number of tillers per unit area about 70 d after transplanting. T_pt and T_pd were about (26±3) d and (54±3) d after transplantation, respectively. An outstanding trend was achieved in the response of the tillering characteristic index to planting density. Except for T_pt and T_sd, the tillering characteristic index followed the power function. N_g, R_it, N_r, T_ed, D_d, R_mt, and R_md showed a power function increase with the increase of planting density, and the exponent of the power function was less than 1, indicating the slow-down trend. T_pt, R_id, T_st, T_et, and D_t also decreased as a power function with the increase of planting density. The exponent of the power function was greater than -1, that is, the decreasing rate gradually decreased with the increase in planting density. N_d increased as a power function with the increase of planting density, where the exponent of the power function was greater than 1, indicating the accelerated increase. T_pd and T_sd decreased with the increase of planting density and then increased gradually after reaching the minimum. (3) A better prediction was achieved, where the R² between the observed and simulated values were 0.96. Therefore, the model can accurately describe the evolution in the number of rice tillers, indicating the better goodness of fitting, adaptability, and interpretability. The model can be applied to the dynamic regularity of the tiller number under the genotypic varieties and the agronomic measures. The tillering dynamic indicators can be expected to serve as the important phenotype in the interaction between genes and environments.