Year: 2026
Type: book-chapter
Abstract: Contaminating crop diseases are serious issues affecting agricultural output.So, the dynamic identification of these problems accurately in a rapid manner is a potential.Traditional disease observation relies on hands-on visual inspection of plant tissue, an approach that is slow and relies heavily on expert advice.That puts it beyond the reach of small farmers.The revolution in deep learning has the potential to enable the automation of the diagnostic process using computational picture analysis.This paper presents the design, building and evaluation of a deep learning system to verify illnesses under investigation in plant leaf images.87K RGB leaf images from 38 labelled classes with healthy and ill specimens during training.The input images are standardized to 240 X 240 pixels and pixel normalization is performed.In addition, augmentation operations including rotation, mirror flipping, random zoom and luminance modification are used to enhance diverse training in order to improve generalization and reduce overfitting.Initially, a baseline CNN is developed to set up a reference performance stage and then three pretrained architectures for allocation learning are integrated.EfficientNetB0, MobileNetV2, and ResNet50 are pretrained on large-scale ImageNet data and are efficient in the plant pathology field.All models are evaluated using accuracy, confusion matrix visualization, and the loss function.EfficientNetB0 generates the prospective disease classification of outcomes, whereas MobileNetV2 is distinguished for its incredibly low inference latency and moderate computing footprint.This research study covers a wide range of potential methods in which intelligent computational models might contribute to modernizing agricultural health monitoring and ensuring long-term food production security.
Authors: Tanisha Ranjan, Onima Tigga, Jaya Pal
Institution: Birla Institute of Technology, Mesra
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: hybrid
OpenAlex ID: W7214933485
