Agri Vision: Deep Learning Based Plant Disease Detection Using Swin Transformer
D. Anil,
Yenduru Sai Jaswanth,
Parakandla Bharath Kumar,
Rajanala Naga Surendra,
Vishnumolakala Venkat Sai Ram
Plant disease detection plays a major role in preci-sion agriculture because delayed or inaccurate
diagnosis directly affects crop quality and productivity. Manual inspection of leaves by farmers or
agricultural experts is time-consuming, subjective, and difficult to scale for large cultivation areas.
This paper presents Agri Vision, a deep learning based plant disease detection system that uses a pretrained
Swin Trans-former model for classifying plant leaf images into healthy or diseased categories.
The proposed pipeline includes dataset loading, image preprocessing, augmentation, transfer learning,
transformer-based feature learning, evaluation, and deployment through a web interface. In contrast
to conventional machine learning approaches that rely on handcrafted features and CNN-only systems
that primarily emphasize local patterns, the Swin Transformer can model both local and global
characteristics of leaf symptoms. The overall framework supports fast, automated, and reliable disease
prediction for smart agriculture applications.
