Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SMARD: A Cost Effective Smart Agro Development Technology for Crops Disease Classification

Tanoy Debnath, Shadman Wadith, Anichur Rahman

TL;DR

SMARD addresses the need for accessible crop disease diagnosis and integrated agritech services for smallholder farmers. The authors design an end-to-end pipeline that localizes leaves with YOLOv5, extracts discriminative features via a CNN with metric learning, and classifies diseases, trained on PlantVillage data, within a web-based platform that also offers advisory and financial services. The system achieves 97.3% classification accuracy and 96% F1-score, outperforming existing agricultural web apps, demonstrating strong potential to improve yields and livelihoods in Bangladesh. This work advances practical, data-driven farmer support by combining computer vision disease detection with extension services and credit access.

Abstract

Agriculture has a significant role in a country's economy. The "SMARD" project aims to strengthen the country's agricultural sector by giving farmers with the information and tools they need to solve common difficulties and increase productivity. The project provides farmers with information on crop care, seed selection, and disease management best practices, as well as access to tools for recognizing and treating crop diseases. Farmers can also contact the expert panel through text message, voice call, or video call to purchase fertilizer, seeds, and pesticides at low prices, as well as secure bank loans. The project's goal is to empower farmers and rural communities by providing them with the resources they need to increase crop yields. Additionally, the "SMARD" will not only help farmers and rural communities live better lives, but it will also have a good effect on the economy of the nation. Farmers are now able to recognize plant illnesses more quickly because of the application of machine learning techniques based on image processing categorization. Our experiments' results show that our system "SMARD" outperforms the cutting-edge web applications by attaining 97.3% classification accuracy and 96% F1-score in crop disease classification. Overall, our project is an important endeavor for the nation's agricultural sector because its main goal is to give farmers the information, resources, and tools they need to increase crop yields, improve economic outcomes, and improve livelihoods.

SMARD: A Cost Effective Smart Agro Development Technology for Crops Disease Classification

TL;DR

SMARD addresses the need for accessible crop disease diagnosis and integrated agritech services for smallholder farmers. The authors design an end-to-end pipeline that localizes leaves with YOLOv5, extracts discriminative features via a CNN with metric learning, and classifies diseases, trained on PlantVillage data, within a web-based platform that also offers advisory and financial services. The system achieves 97.3% classification accuracy and 96% F1-score, outperforming existing agricultural web apps, demonstrating strong potential to improve yields and livelihoods in Bangladesh. This work advances practical, data-driven farmer support by combining computer vision disease detection with extension services and credit access.

Abstract

Agriculture has a significant role in a country's economy. The "SMARD" project aims to strengthen the country's agricultural sector by giving farmers with the information and tools they need to solve common difficulties and increase productivity. The project provides farmers with information on crop care, seed selection, and disease management best practices, as well as access to tools for recognizing and treating crop diseases. Farmers can also contact the expert panel through text message, voice call, or video call to purchase fertilizer, seeds, and pesticides at low prices, as well as secure bank loans. The project's goal is to empower farmers and rural communities by providing them with the resources they need to increase crop yields. Additionally, the "SMARD" will not only help farmers and rural communities live better lives, but it will also have a good effect on the economy of the nation. Farmers are now able to recognize plant illnesses more quickly because of the application of machine learning techniques based on image processing categorization. Our experiments' results show that our system "SMARD" outperforms the cutting-edge web applications by attaining 97.3% classification accuracy and 96% F1-score in crop disease classification. Overall, our project is an important endeavor for the nation's agricultural sector because its main goal is to give farmers the information, resources, and tools they need to increase crop yields, improve economic outcomes, and improve livelihoods.
Paper Structure (11 sections, 4 figures)

This paper contains 11 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Methodology
  4. Leaf Object Detection
  5. Dataset Collection
  6. Feature Extraction and Classification
  7. Network Architecture
  8. Proposed System
  9. Performance Analysis and Discussion
  10. Limitations and Future Scope
  11. Conclusion

Figures (4)

  • Figure 1: PlantVillage dataset
  • Figure 2: Crops Disease Detection Pipeline of "SMARD"
  • Figure 3: Implementation of Plant disease classification using pre-trained CNN Network
  • Figure 4: Proposed workflow of our web application based system "SMARD"