🔬 CSE-AIML Research Prototype

AI-Powered Crop Disease Detection & Smart Agriculture

“Helping agricultural scientists and farmers detect microscopic crop pathogens earlier using Computer Vision and Machine Learning.”

Explore the Project ↓ Start AI Demo →

📷 Computer Vision: ACTIVE

🦠 Disease Detection: SCANNING

🧠 AI Model: MobileNetV3

📊 Crop Health: 98% (Normal)

Problem Statement

🌾 The Challenge Scientists Face

Agricultural scientists and farmers face major hurdles in protecting food security. Manual inspection is slow and diseases are often noticed only after severe crop damage has occurred.

🦠

Crop Diseases

Pathogens like Early Blight and Late Blight spread rapidly across acres before symptoms are clearly visible to human scouts.

Global Loss~40% Yield

🌡️

Heat Stress

Sudden temperature spikes cause rapid water evapotranspiration and leaf necrosis, stunting crop development.

VulnerabilityHigh Heat Risk

💧

Water Scarcity

Fixed-schedule watering causes severe resource waste or under-irrigation. Real-time soil monitoring is essential.

Water WastedUp to 60%

🐛

Pest Attacks

Manual scouting of vast agricultural fields is labor-intensive, slow, and prone to subjective human error.

Inspection Delay5-7 Days

🌦️

Changing Climate

Unpredictable rainfall patterns and microclimate shifts make traditional seasonal farming predictions unreliable.

Planning RiskHigh Volatility

Interactive 3D Simulation

🔬 3D Scientific Visualization: Disease Progression

Move the slider below to observe how plant leaves physiologically change from healthy tissue to disease-infected lesions, and see how Computer Vision detects the affected areas.

🟢 Healthy Tissue 🔴 Diseased Foliage

Slide to simulate Early Blight fungal development

🌿

Leaf Condition: Optimal & Vigorous

🔍

AI Detection Area: 0% (No Pathogens Detected)

💡

Scientific Note: Cellular chlorophyll levels optimal. Normal photosynthesis active.

System Pipeline

🧠 How AI Helps: The Diagnostic Pipeline

AgriAI transforms agricultural monitoring from slow, reactive manual scouting into an instant, data-driven, predictive decision system.

STEP 1

📷

Plant Image

Camera or drone captures high-res leaf macro-photography in the field.

STEP 2

👁️

Computer Vision

Image resizing, color normalisation, and background segmentation.

STEP 3

🧠

AI Model

Deep CNN extracts spatial feature maps representing lesion geometry.

STEP 4

🔍

Detection

Classifies healthy leaves vs specific fungal and viral pathogens.

STEP 5

📊

Confidence

Computes probabilistic certainty score (e.g. 94.2% confidence).

STEP 6

💡

Recommendation

Generates safe agronomic advisory for scientists and farmers.

Deep Learning Architecture

🧬 Inside the Neural Network

How Convolutional Neural Networks (CNNs) process raw pixel matrices through mathematical filter kernels to identify microscopic disease patterns.

Input Layer: Raw 224×224 RGB Leaf Pixels → Hidden Conv Layers: Edge & Texture Extractors → Output Layer: Softmax Probabilities

* Conceptual visualization of the AI classification process.

Interactive Laboratory Demo

🔬 Live AI Crop Disease Detection

Choose a curated leaf specimen below and click "Analyze Leaf" to trigger the real-time AI scanning sequence.

Target Crop: Tomato (Solanum lycopersicum)

Detected Condition: Early Blight (Alternaria solani)

Pathogen Classification: Fungal Infection

AI Confidence Score: 94.2%

Risk Severity: HIGH (Immediate Attention)

AI Recommendation: Concentric brown target spots identified on lower foliage. Recommend pruning affected leaves, improving row ventilation, and consulting an agricultural advisor.

⚠️ Important Note: This is an educational research prototype designed for demonstration, not a certified agronomic prescription system.

IoT Extension

⚡ Connecting Vision with IoT Sensors

In an advanced smart farm, optical disease detection is fused with real-time environmental telemetry to automate smart irrigation and climate control.

🌱

Plant Foliage

HD Camera monitors visual leaf patterns continuously.

🌡️

Sensors

DHT22 temp/humidity + Capacitive soil moisture probe.

⚡

ESP32 Edge MCU

Streams wireless telemetry via MQTT to AI cloud engine.

🧠

AgriAI Engine

Fuses vision + telemetry to evaluate total stress risk.

💧

Smart Irrigation

Actuates 5V relay to hydrate soil only when required.

Impact & Benefits

🚜 Why AI Can Help Agriculture

Empowering agricultural scientists and farmers with fast, objective, automated diagnostic intelligence.

120 ms

⚡ Faster Monitoring

Computer Vision analyzes hundreds of crop images in milliseconds compared to days of manual field walking.

5-7 Days

🔍 Early Detection

Identifies microscopic fungal patterns days before widespread leaf chlorosis causes irreversible yield loss.

Multi-Modal

📊 Data-Driven Decisions

Integrates optical leaf patterns with temperature, humidity, and soil moisture telemetry.

38%

💧 Water Saved

Automated closed-loop irrigation triggers only when soil moisture drops below critical physiological thresholds.

Human-AI Collaboration

🔬 From Scientist → AI Assistant

Understanding the philosophical role of Artificial Intelligence in agricultural science.

Traditional Approach

Manual Field Scouting → Visual Observation → Delayed Disease Identification → Reactive Chemical Spraying

AI-Assisted Approach

Optical Capture → Instant AI Pattern Analysis → Risk Prediction & Confidence → Scientist's Informed Decision

“AI assists agricultural scientists; it does not replace them.”

Future Roadmap

🚀 The Future of AgriAI

Scaling from single-leaf diagnostics to farm-scale autonomous agricultural intelligence.

PHASE 1

📷 Leaf Diagnostics

MobileNet CNN classification of tomato leaf diseases.

PHASE 2

🌡️ Stress Prediction

Random Forest ML combining temp, humidity, and moisture.

PHASE 3

💧 Smart Irrigation

ESP32 microcontrollers driving physical water pumps.

PHASE 4

🚁 Drone Monitoring

Autonomous UAVs conducting aerial farm NDVI scans.

PHASE 5

🛰️ Satellite GIS

Multispectral satellite tracking of regional crop health.

PHASE 6

🤖 Autonomous Farming

Edge TinyML robotics for precision field management.

Architecture Diagram

📐 Clickable System Architecture

Click on any module below to understand its role in the end-to-end pipeline.

🌱 Crop

→

📷 Camera

→

⚙️ Processing

→

👁️ Vision

→

🧠 AI Model

→

🔍 Prediction

→

📊 Risk Analysis

→

💡 Advisory

→

💻 Dashboard

Click any block above to view details.

Capstone Presentation

Presented By

Adithya

CSE – Artificial Intelligence & Machine Learning

Project: AgriAI • College Exhibition & Defense

“Hi! I’m Adithya AI. I’ll explain how this project uses Artificial Intelligence and Computer Vision to detect diseases in crops. Choose a topic below!”

💡 Simple Terminology:

• CNN: Neural network designed to recognize image shapes.

• Confidence: How strongly the AI supports its prediction.

🌱

AgriAI • Presentation Mode

Presenter: Adithya D (CSE-AIML)

Slide 1 of 8

INTRODUCTION

AgriAI: Smart Crop Health

🤖 Adithya AI Spoken Script:

“Hello everyone. Our project is called AgriAI...”