Object detection models have become the current tool of choice for plant disease detection in precision agriculture. Most existing research improved the performance by ameliorating networks and optimizing the loss function. However, because of the vast influence of data annotation quality and the cost of annotation, the data-centric part of a project also needs more investigation. We should further consider the relationship between data annotation strategies, annotation quality, and the model’s performance. In this paper, a systematic strategy with four annotation strategies for plant disease detection is proposed: local, semi-global, global, and symptom-adaptive annotation. Labels with different annotation strategies will result in distinct models’ performance, and their contrasts are remarkable. An interpretability study of the annotation strategy is conducted by using class activation maps. In addition, we define five types of inconsistencies in the annotation process and investigate the severity of the impact of inconsistent labels on model’s performance. Finally, we discuss the problem of label inconsistency during data augmentation. Overall, this data-centric quantitative analysis helps us to understand the significance of annotation strategies, which provides practitioners a way to obtain higher performance and reduce annotation costs on plant disease detection. Our work encourages researchers to pay more attention to annotation consistency and the essential issues of annotation strategy. The code will be released at: https://github.com/JiuqingDong/PlantDiseaseDetection_Yolov5 .

This study aimed to determine the optimal indole-3-acetic acid (IAA) concentration in a nutrient solution to increase the bioactive compounds while enhancing the plant growth of A. rugosa grown hydroponically. Twenty-eight-day-old plants were transplanted in a plant factory for 32 days. The plants were subjected to various IAA concentrations (10−11, 10−9, 10−7, and 10−5 M) from 8 days after transplanting, and the control treatment (without IAA). Shoot and root fresh weights were effectively improved under 10−7 and 10−9 IAA treatments. Leaf gas exchange parameters were increased under 10−7 and 10−9 IAA treatments. Four of the IAA treatments, except 10−11 IAA treatment, significantly increased the rosmarinic acid (RA) concentration, as well as the tilianin concentration was significantly increased at all IAA treatments, compared with that of the control. Especially, the tilianin concentration of the 10−11 IAA treatment was significantly (1.8 times) higher than that of the control. The IAA treatments at 10−5 and 10−7 significantly raised the acacetin concentrations (1.6- and 1.7-times, respectively) compared to those of the control. These results suggested that 10−7 concentration of IAA in a nutrient solution was effective for enhancing plant growth and increasing bioactive compounds in A. rugosa, which offers an effective strategy for increasing phytochemical production in a plant factory.

Recognizing plant diseases is a major challenge in agriculture, and recent works based on deep learning have shown high efficiency in addressing problems directly related to this area. Nonetheless, weak performance has been observed when a model trained on a particular dataset is evaluated in new greenhouse environments. Therefore, in this work, we take a step towards these issues and present a strategy to improve model accuracy by applying techniques that can help refine the model’s generalization capability to deal with complex changes in new greenhouse environments. We propose a paradigm called “control to target classes.” The core of our approach is to train and validate a deep learning-based detector using target and control classes on images collected in various greenhouses. Then, we apply the generated features for testing the inference of the system on data from new greenhouse conditions where the goal is to detect target classes exclusively. Therefore, by having explicit control over inter- and intra-class variations, our model can distinguish data variations that make the system more robust when applied to new scenarios. Experiments demonstrate the effectiveness and efficiency of the proposed approach on our extended tomato plant diseases dataset with 14 classes, from which 5 are target classes and the rest are control classes. Our detector achieves a recognition rate of target classes of 93.37% mean average precision on the inference dataset. Finally, we believe that our study offers valuable guidelines for researchers working in plant disease recognition with complex input data.

This study was conducted to examine the changes of photosynthesis, growth, chlorophyll contents and functional material contents in light intensity and EC concentration of wild baby leaf vegetable, Indian lettuce (Lactuca indica L. cv. 'Sunhyang') in DFT hydroponics. The cultivation environment was 25±1°C of temperature and 60±5% of relative humidity in growth system. At 14 days after sowing, combination effect of light intensity (Photosynthetic Photon Flux Density (PPFD 100, 250, 500 μmol•m -2 •s-1 ) and EC level (EC 0.8, 1.4, 2.0 dS․m -1 ) of nutrient solution was determined at the baby leaf stage. The photosynthesis rate, stomatal conductance, transpiration rate and water use efficiency of Indian lettuce increased as the light intensity increased. The photosynthesis rate and water use efficiency were highest in PPFD 500-EC 1.4 and PPFD 500-EC 2.0 treatment. The chlorophyll content decreased as the light intensity increased, but chlorophyll a/b ratio increased. Leaf water content and specific leaf area decreased as light intensity increased and a negative correlation (p < 0.001) was recognized. Plant height was the longest in PPFD 100-EC 0.8 and leaf number, fresh weight and dry weight were the highest in PPFD 500-EC 2.0. Anthocyanin and total phenolic compounds were the highest in PPFD 500-EC 1.4 and 2.0 treatment, and antioxidant scavenging ability (DPPH) was high in PPFD 250 and 500 treatments. Considering the growth and functional material contents, the proper light intensity and EC level for hydroponic cultivation of Indian lettuce is PPFD 500-EC 2.0, and PPFD 100 and 250, which are low light conditions, EC 0.8 is suitable for growth.

No abstract