Early detection and classification
of banana leaf disease using image
processing techniques
Team members
711721104083 – Poorani R
711721104094 – Sabitha K
711721104103 – Shapna G
711721104105 – Sivaramakrishnan D
Guided By
Ms. Suganthi A, AP/CSE
Objective
The objective of this project is to develop an early detection system for banana leaf diseases using image
processing techniques. The system aims to help farmers identify diseases in the initial stages, allowing them to
take timely action and prevent major crop losses. By integrating a Raspberry Pi camera, the system captures and
analyzes leaf images in real-time, classifies diseases using machine learning algorithms .To achieve accurate
disease classification, the project utilizes Convolutional Neural Networks (CNNs), specifically ResNet50 and
InceptionV3 architectures, to extract key features such as color, texture, and shape patterns from banana leaves.
Abstract
Banana is one of the most important fruit crops in India, but its yield is significantly affected by various leaf
diseases. Traditional disease detection methods are manual, time-consuming, and often inaccurate, leading to
delayed treatment and reduced productivity. This project, "Early Detection and Classification of Banana Leaf
Disease Using Image Processing Techniques," aims to provide an automated real-time detection system using
deep learning. We use CNN, ResNet, and InceptionV3 models to analyze banana leaf images and identify disease
symptoms early. A Raspberry Pi camera or mobile device captures leaf images, which are processed to detect
infections based on texture, color, and shape features. Early detection helps farmers prevent disease spread,
optimize potassium uptake, and improve overall crop health. This system reduces excessive pesticide use, ensuring
better fruit quality and sustainable farming. By offering a cost-effective AI-driven solution, it makes disease
detection accessible to all farmers. The project ultimately contributes to higher banana yield, increased profits,
and improved food security in India.
Existing System
The study developed a Convolutional Neural Network (CNN)-based model to detect plant diseases using
images from the Plant Village dataset.
Used machine learning (SVM, K-means, GLCM) to detect potato leaf diseases.
Pre-trained models (like VGG, ResNet) improved accuracy but required high computational power.
Proposed System
A Raspberry Pi camera captures banana leaf images, enabling real-time data on disease detection
The system uses CNN, ResNet50, and InceptionV3 models to detects early-stage infections, and to
analyze leaf texture, color, and shape, ensuring precise identification of diseases
Outcome
This project integrates CNN models like ResNet50 and InceptionV3, the system provides highly accurate
classification of banana leaf diseases in real-time. The use of a Raspberry Pi camera allows farmers to easily
capture leaf images and receive instant disease diagnosis with the use of real time data. It enhances potassium
uptake efficiency, ensuring healthier crops and better fruit quality. With early detection, farmers can prevent the
spread of infections, apply targeted treatments, and reduce pesticide overuse, leading to healthier crops and
increased yield. Ultimately, this project contributes to sustainable agriculture, improved food security, and better
farm management.
Innovation Component to be Derived
Integrates Raspberry Pi camera for on-field real-time disease identification, enabling instant diagnosis
for farmers.
Uses Convolutional Neural Networks (CNNs) like ResNet50 and InceptionV3 to accurately classify
banana leaf diseases based on color, texture, and shape patterns.
Detects symptoms at an early stage using real-time image data, allowing farmers to take preventive
actions before severe infection spreads. By detecting diseases before major outbreaks, farmers can
apply targeted treatments instead of excessive pesticide use, ensuring cost-effective and eco-friendly
farming.
Works with affordable hardware (Raspberry Pi + simple camera), making it accessible for small-scale
farmers and ensuring sustainable agriculture.
Helps optimize crop management by preventing late-stage infections, leading to higher yield, improved
food security, and reduced losses.
Timeline
Particulars Feb Mar Apr May
1 2 1 2 3 4 1 2 3 4 1 2 3 4
Literature review
Camera module
Raspberry component
Train model
Implement image
preprocessing
Developing the user
interface
Integrate trained models
with Raspberry Pi
Refine disease
classification
Improve detection speed
Deploy the final system