Advanced Guide to Cotton Leaf Area Prediction: How to Apply Artificial Intelligence
Author: Cheng Weilun
This article will give everyone a glimpse of how agriculture combined with artificial intelligence can use big data to predict an important factor affecting cotton yield: the leaves that carry out photosynthesis.
- Stage 1 [A brief discussion of the methods of use and an overview]
- Stage 2 [Let us talk about cotton leaf area and how we measured it in the past]
- Stage 3 [How do human brain neurons become artificial neural networks?]
- Stage 4 [Input data | Sample collection of cotton leaves]
- Stage 5 [After training the artificial intelligence, what answers will we get? Further analysis]
______________________________________
Stage 1 [A brief discussion of the methods of use and an overview]
Everyone knows that plants produce energy through photosynthesis in their leaves, so the size of the leaf area becomes an important key to whether a crop grows well and to its productivity. Therefore, scientists conducted experiments to use mathematical prediction, paired with different instruments, to estimate the leaf area of vegetable and fruit crops.
Figure. The relative importance of the four input nodes to the network response
In research on crop growth and productivity, leaf area is an important indicator, representing plant growth, estimating plant productivity, analyzing nutrient uptake, water management based on leaf area, weed management and other organisms harmful to plants. Therefore, different instruments and prediction models have been used and developed to estimate the leaf area of different crops, such as vegetables and fruits.
In this study, an artificial neural network (ANN) predicted cotton leaf area. The artificial neural network model was able to explain 97.54 percent of the variation in the leaf area of the calibration (training) data and 92.32 percent of the variation in the validation (testing) data. Leaf width, leaf length, the length of the second right leaf and the second left leaf length were used as input variables respectively. Although leaf width, leaf length, the length of the right second leaf and the second left leaf length all significantly contributed to the prediction of the artificial neural network model, the left leaf length was slightly better, contributing 32.58 percent of the predictive power.
Tools help us understand and observe this world: for microorganisms that are hard to see with the naked eye, we use a microscope; when the naked eye cannot see far, we use binoculars; when we cannot swim such a long distance, there are boats; when we cannot dive, there are submarines. And amid subtle distinctions, or in the vast sea of data, things that humans find hard to remember and hard to generalize and judge, for example, whether height being taller is usually influenced more by the right hand or the left hand, we can use large amounts of data, artificial intelligence and neural networks, and the deep learning of computers, to extend our five senses.
But as for the final results, what meaning to give them, and how to judge the data results, still depends on us humans.
Just like using a microscope to help us see microorganisms: we are not afraid of the microscope, since after all it is only a tool. And after seeing the microorganisms, what use is that? Only we ourselves know.
Publication: International Journal of Plant & Soil Science 8(4): 1-13, 2015; Article no.IJPSS.19686
Authors: A. M. Aboukarima, H. A. Elsoury and M. Menyawi