GRENZE International Journal of Engineering and Technology
Vol. 12
(2026), Issue 2
An Image-based Study on Crop Disease Detection
Authors
Mohammad Areeb, Ayush Tripathi, Aman Bohra, Yash Vinayak Singh, Rahul Kumar Sharma, Devendra Gautam
Abstract
Crop diseases is a major problem for farmers all around the world as it results in reduced yield and losses in billions of dollars. To contribute in solving this problem we developed an automated crop disease detection system which can detect diseases by analyzing a photograph of a leaf. We have implemented a deep learning-based system to classify crop diseases. We have trained a Convolutional Neural Network (CNN) based model on PlantVillage Dataset containing a total of 61,486 images across 38 classes (each class corresponds to a type of disease with a class for healthy leaves as well). We have pre-processed and augmented the data before training it on the model. This is done to improve the accuracy of our model in real world scenarios. Steps like resizing the images to a fixed size, cropping the images, rotating the images, varying brightness etc. were implemented as part of preprocessing and augmentation steps. The model is then used to visually analyse the diseased portions on the leaf and display the results quickly on a user interface. Model has a classification accuracy of 98.7% for images under control conditions as of initial testing (accuracy may vary with other factors like environment, background, lighting, quality of image etc.). Currently, the system is built using the CNN model. The plan is to build and train more complex and efficient models in the future using models like EfficientNet, ResNet, Vision Transformers and more. We also plan to use the trained model for a mobile application in the future to identify plant diseases in the field. The proposed system is cost-effective, fast, easy-to-use, and highly scalable which can potentially help in reducing crop losses and aiding farmers.
Pages:
1998 - 2003