SIDDHARTH INSTITUTE OF ENGINEERING & TECHNOLOGY
(AUTONOMOUS)
DEPARTMENT OF ELECTRONICS AND COMMUNICATIONS ENGINEERING
SECTION NO. -2 BATCH NO -12
Ensemble Deep Learning Framework for Multiclass
Plant Disease and Nutrient Deficiency Detection
TEAM MEMBERS: Under the esteemed guidance of :
[Link]
NELLORE LAKSHMIPRASANNA - 22F61A04A4
ASSISTANT PROFESSOR
NAGISETTI KALYANI - 22F61A0494 ECE,SIETK
SEKHAR GANGADRI - 22F61A0465
NELLORE LIKHITHA - 22F61A04B0
NARUBOYINA HARSHA VARDHAN - 22F61A0481
ABSTRACT
Growing healthy plants is essential for sustainable agriculture and food
production. Due to their comparable visual symptoms and environmental
impacts, plant diseases and nutritional deficits are difficult to recognize. This
ensemble deep learning model uses leaf pictures to detect and categorize plant
illnesses and nutrient deficits. The suggested system takes advantage of
VGG16 and MobileNetV2 deep learning models. VGG16 extracts deep visual
features, while MobileNetV2 processes faster with light weight architecture.
They create a balanced, accurate real-time analytical model. The model
was trained on a broad dataset of plant leaf images and preprocessed with
image scaling, normalization, and augmentation to reduce overfitting. The
ensemble model accurately detected numerous disease types and nutritional
concerns, making it ideal for agriculture. Early diagnosis helps farmers
prevent crop loss and boost output. Modern precision farming is affordable
and easy with the system on mobile devices or smart agricultural platforms.
Motivation Behind The Project
i. In real agricultural fields, nutrient deficiencies and diseases often show similar visible
symptoms, making it extremely difficult for farmers to diagnose the exact problem without
expert help.
ii. Using wrong treatment such as unnecessary pesticides or excess fertilizers leads to crop
loss, soil damage, and economic losses.
iii. Existing systems mainly detect either diseases or nutrient deficiencies, but not both together.
iv. There is no integrated, field-deployable system capable of handling both tasks in real time.
v. This project is motivated by this practical gap. By building a combined deep learning
ensemble system, the project aims to:
* Provide accurate diagnosis even when symptoms overlap
*Help farmers take correct decisions quickly
*Reduce dependency on experts
*Enable mobile-based, on-field plant health analysis
*Improve crop yield and support sustainable agriculture
Objectives
• To develop an automated ensemble deep learning model that can
identify and classify plant leaf diseases and nutrient deficiencies.
• To enhance classification accuracy by combining multiple deep
learning models (VGG16 and MobileNetV2) through ensemble
techniques.
• To build a practical and scalable solution for farmers and agronomists,
enabling fast diagnosis through visual inspection using leaf images.
Applications
Smart Agriculture / Precision Farming:
*Early detection of plant diseases and nutrient deficiencies
*Reduction in unnecessary pesticide and fertilizer usage
Mobile & Edge-Based Field Diagnosis:
*Can be deployed on smartphones, Raspberry Pi, drones
Agricultural Advisory Systems:
*Helps farmers make informed decisions quickly
Crop Monitoring Systems:
*Continuous monitoring using AI-based farm cameras
Research & Agricultural Extension Services
*Useful for agronomists, researchers, and government agencies
Thank you