Plant diseases have plagued agriculturalists and farmers alike since at least 752 AD (Saunders, 2003). Many have suffered through countless seasons of poverty and famine because of crop-destroying diseases. With the onset of the modern age of technology, researchers have started to find new ways to enhance disease detection that are easy for farmers and agriculturalists to use to try and catch plant diseases early on and predict spread in order to prevent mass plant casualties.

Molecular approaches used to detect plant diseases include PCR tests to detect pathogens inside plant leaves. They “use the total DNA from the diseased plant tissue as a template to detect the target organisms” (Grosdidier, 2016). Other methods include using nanopores to diagnose plant diseases, where single-molecule sequencing technology only requires a small amount of DNA to sequence the nucleic acid in plants with diseases. Then, plants can be diagnosed when researchers compare their nucleic acid with the nucleic acids of diseased plants (Chalupowicz, 2018). However, researchers have found drawbacks to molecular approaches to detecting plant diseases. In the research paper “Are molecular tools solving the challenges posed by detection of plant pathogenic bacteria and viruses?” by Maria Lopez, she and her team write, “there are some unsolved problems concerning the detection of many plant pathogens due to their low titre in the plants, their uneven distribution, the existence of latent infections and the lack of validated sampling protocols” (Lopez, 2009). Lopez’s team raises questions about whether PCR detection is the best way to detect plant pathogens that are in lower concentrations in plants, especially in the current era with more advanced technology. Furthermore, it isn’t easy for agriculturists to identify exactly what disease their crops are afflicted with without sending samples to a lab, which could take a while to get results back. Many agriculturalists are also not well versed in molecular-based techniques to run these tests themselves. Thus, there must be a better and faster way for them to catch diseases in their crops.

Outside of the molecular approach, artificial intelligence machine learning is one of the most prominent techniques for discovering plant diseases. Researchers are using “a robust drone-based deep learning approach,” where they have used EfficientNetV2-B4, a convolutional neural network that takes key data points from plants with certain diseases and classifies them into classes. Here, they use samples taken by a drone to detect diseases. This methodology has allowed them to reach “average precision, recall, and accuracy values of 99.63%, 99.93%, and 99.99%, respectively (Albattah, 2022). Other machine-learning techniques include training artificial intelligence under a Random Forest classification method to sort images of plant leaves into diseased and healthy. Overall, “the objective of this algorithm is to recognize abnormalities that occur on plants in their greenhouses or natural environment” (Ramesh, 2018). Once technology like this is commercialized into easily accessible apps or handheld products, plant diseases can be caught in the early stages before massively spreading across counties. For agriculturists, this is a huge step up from simply using visual recognition of diseases in their crops or sending their plant samples to labs, as using AI is a quicker and more efficient way to do so as long as the accuracy of the AI is high and the software is trustworthy.

Not only can AI be used to detect plant diseases, but it can also predict how diseases will spread. Much like the neural networks mentioned before, researchers have programmed algorithms to extract data on the weather and pest appearance to forecast the spread of diseases that have been detected in a certain area (Domingues, 2022). Domingues and his team tested this technology on diseases such as the Tomato Powdery Mildew Disease and found that once the AI was trained with images from the field rather than the laboratory, the accuracy was boosted immensely. Furthermore, The Symbiosis Institute of Technology found that Artificial Neural Networks outperformed numerous other algorithms they tested (Support Vector Machines, K-Nearest Neighbors…etc.) (Patil, 2022). John Quinn and his team confirm that this technology would be easy for “workers with basic training” to use as the “diagnosis of plant disease can be automated using images taken by a camera phone” as the detection method achieved an AUC value of 0.959-0.961 (Quinn, 2011). Thus, this is another reason why this technology would be easy to use for agriculturists while also being immensely accurate.

Though using a molecular approach is applicable in contexts regarding research and initial disease identification, on the field, using Artificial Intelligence with visual recognition and machine learning will be better suited to detect crop diseases on the field as surveyists and farmers can use technology in a handheld device or their phone to get information on the health of their crops. Quick detection methods through technology will also decrease the use of harmful pesticides used to kill plant virus vectors (Karpyshev, 2021). This streamlines disease detection and helps reduce large unintentional environmental impacts when trying to prevent disease. All in all, the ability to predict the spread of plant diseases and identify diseases without sending samples to the lab is a huge revolution in disease prevention for workers who make a living off crops, especially in third-world countries.

Works Cited:

Albattah W, Javed A, Nawaz M, Masood M, Albahli S. Artificial Intelligence-Based Drone System for Multiclass Plant Disease Detection Using an Improved Efficient Convolutional Neural Network. Front Plant Sci. 2022 Jun 9;13:808380. doi: 10.3389/fpls.2022.808380. PMID: 35755664; PMCID: PMC9218756.

Chalupowicz, L., Dombrovsky, A., Gaba, V., Luria, N., Reuven, M., Beerman, A., Lachman, O., Dror, O., Nissan, G. and Manulis-Sasson, S. (2019), Diagnosis of plant diseases using the Nanopore sequencing platform. Plant Pathol, 68: 229-238. https://doi.org/10.1111/ppa.12957

Domingues, T.; Brandão, T.; Ferreira, J.C. Machine Learning for Detection and Prediction of Crop Diseases and Pests: A Comprehensive Survey. Agriculture 2022, 12, 1350. https://doi.org/10.3390/agriculture12091350

Grosdidier, M., Aguayo, J., Marçais, B. and Ioos, R. (2017), Detection of plant pathogens using real-time PCR: how reliable are late Ct values?. Plant Pathol, 66: 359-367. https://doi.org/10.1111/ppa.12591

López MM, Llop P, Olmos A, Marco-Noales E, Cambra M, Bertolini E. Are molecular tools solving the challenges posed by detection of plant pathogenic bacteria and viruses? Curr Issues Mol Biol. 2009;11(1):13-46. Epub 2008 Jun 25. PMID: 18577779.

Patil, R.R., Kumar, S., Rani, R. (2022). Comparison of artificial intelligence algorithms in plant disease prediction. Revue d’Intelligence Artificielle, Vol. 36, No. 2, pp. 185-193. https://doi.org/10.18280/ria.360202

P. Karpyshev, V. Ilin, I. Kalinov, A. Petrovsky and D. Tsetserukou, “Autonomous Mobile Robot for Apple Plant Disease Detection based on CNN and Multi-Spectral Vision System,” 2021 IEEE/SICE International Symposium on System Integration (SII), 2021, pp. 157-162, doi: 10.1109/IEEECONF49454.2021.9382649.

Quinn, John Alexander, Kevin Leyton-Brown, and Ernest Mwebaze. “Modeling and monitoring crop disease in developing countries.” Twenty-Fifth AAAI Conference on Artificial Intelligence. 2011. Saunders, K., Bedford, I., Yahara, T. et al. The earliest recorded plant virus disease. Nature 422, 831 (2003). https://doi.org/10.1038/422831a

S. Ramesh et al., “Plant Disease Detection Using Machine Learning,” 2018 International Conference on Design Innovations for 3Cs Compute Communicate Control (ICDI3C), 2018, pp. 41-45, doi: 10.1109/ICDI3C.2018.00017.