Sri Lanka-focused AI + IoT system

Turn Soil Data Into Confident Crop Decisions

AgriVora combines soil image analysis, real-time pH sensing, GPS, and weather data to deliver ranked crop recommendations with suitability scores designed specifically for Sri Lankan farmers.

View Live Demo Try the System

95%

Prediction Accuracy

50+

Crops Supported

1k+

Trusted Prototype Users

pH Measured

Optimal: 6.5

GPS + Weather

Synced successfully

Target Audience

Built for Everyone in Agriculture

Clear, measurable benefits for professionals, learners, and hobbyists — from the field to your backyard.

Why AgriVora?

From Guesswork to Data-Driven Yields

We identified a critical pain point in agriculture and engineered a smart, practical solution.

The Problem

Farmers often rely on trial-and-error because precise soil conditions are difficult to measure in the field. Without reliable insights into soil texture, pH, and local climate factors, crop selection becomes a gamble—leading to preventable crop failure, wasted input costs, and lower yields.

Limited access to lab testing and agronomy support
Uncertainty about exact soil pH and texture before planting
"One-size-fits-all" advice fails against regional variability

The AgriVora Solution

AgriVora brings lab-grade intelligence to a mobile app. We use a Convolutional Neural Network (CNN) for soil image analysis, an ESP32 IoT pH sensor for real-time data, and GPS APIs for weather context. Our LightGBM model then computes exactly which crops will thrive.

Fast, consistent recommendations based on actual field data
Ranked crop lists with transparent suitability scoring
Actionable soil improvement tips to maximize success

The Workflow

How AgriVora Works

Four seamless steps transform raw field data into highly confident crop recommendations.

Capture Image

Advanced CNN classifies soil texture directly from your mobile camera.

Measure pH

Our ESP32 + Gravity pH sensor syncs calibrated readings instantly via Bluetooth.

Add Context

App securely requests GPS and weather APIs to evaluate regional suitability.

Get Ranked Crops

LightGBM engine processes the inputs to yield top crop recommendations & tips.

Tech Platform

Powerful Features. Smart Tech Stack.

Designed to be highly accurate for researchers and incredibly practical for farmers.

LightGBM Crop Ranking Model

Our sophisticated backend utilizes LightGBM to provide highly accurate predictions with transparent suitability scores that explain the "why" behind every recommendation.

FastAPI Backend
High Accuracy
50+ Crops

Soil Image CNN

Real-time soil texture classification from mobile camera imagery, tailored for Sri Lankan soils.

ESP32 IoT Integration

Stream live pH data straight into the app. Scalable pipeline built for modern field hardware.

GPS & Weather Integration

Context-aware system. Recommendations adapt dynamically to local weather patterns, humidity, and location-specific agricultural data.

Platform Walkthrough

See AgriVora in Action

Watch exactly how simple it is to scan, sync, and predict.

Product Demo Flow

Click to play video

User Interface

Designed for the Field

A clean, accessible interface built for real farmers operating in high-glare field conditions.

Smart Permissions Setup

Guided, user-friendly onboarding protecting data privacy.

AI Crop Recommendations

Clear rankings, confidence scores, and actionable treatment tips.

Field History & Map View

Access historical soil analyses and geospatial trends over time.

Join The Network

Measurable Outcomes.
Real Credibility.

AgriVora is validated by researchers and trusted in preliminary field trials. We prioritize farmer-friendly adoption with uncompromising academic rigor.

Request Prototype Access

Join researchers and farmers using AgriVora.

Reduce Trial-and-Error

Replace guesswork with scientifically backed models directly synced to your soil environment.

Cost-Effective Testing

Skip expensive, slow lab testing. Generate rapid, reusable insights directly on-site.

Tested & Built For:

Sri Lankan Agriculture

University Research

Questions

Frequently Asked Questions

Yes. AgriVora includes a Manual Mode where you can manually enter known soil texture and pH values. However, pairing with our ESP32 sensor pipeline guarantees highly accurate, localized predictions.

Our LightGBM model boasts an impressive ~95% prediction accuracy on current datasets. It combines CNN visual texture classification with hard numerical data (pH + Climate) to ensure trustworthy output.

Internet access is required to fetch real-time weather APIs and process heavy ML tasks on our FastAPI backend. However, the app can queue local sensor readings offline and sync once connection is restored.

Yes, the current datasets and predictive algorithms have been hyper-optimized for Sri Lankan soil types, regional climates, and local crop varieties to ensure maximized relevance and yield.