NED University Student Venture

AgriVision

"Know when to sell. Know what's wrong with your crop."

A simple advisor for smallholder farmers in Pakistan. AgriVision turns price, weather, and crop data into plain-language sell or hold recommendations and checks leaf photos for early disease detection.

Live Trend Simulation Sample data
SELL

Company Profile & Context

Data Science for Smallholder Agriculture

Student Venture Profile

AgriVision is an early-stage student venture focused on applying data science to smallholder agriculture in Pakistan.

Built specifically to address local agricultural challenges, our objective is to turn fragmented open data into actionable market and agronomic guidance.

Focus: Pakistan Agritech & Smallholder Support

Academic Roots

Founded by students at NED University of Engineering and Technology.

Developed as part of the course "Data Science Platform Architecture and Application". Note: AgriVision is currently an academic project and not a registered business.

Institution: NED University of Engineering & Technology
The Core Problem

"Many farmers lack timely, reliable information on when to sell and how to spot crop disease early."

Smallholder farmers face two critical information gaps during every growing season.

1. Selling Blind

Market Information Gap

Without access to price trend forecasts, farmers often sell immediately after harvest when market supply peaks and prices drop, or hold too long during market declines.

Volatile Market Price Uncertainty High
Sell now or wait?

2. Disease Noticed Too Late

Crop Health Gap

Crop diseases spread rapidly. Without diagnostic tools, early symptoms pass unverified until damage severely compromises yields and farm income.

Visual Symptoms
• Late detection leads to crop loss
• Treatment delayed

Who Faces This & When?

Smallholder farmers across Pakistan face these decisions continuously around harvest planning and throughout the growing season.

Impact: Yield & Income Loss
Our Approach

How AgriVision Addresses The Problem

Turning raw, public data sources into simple advice delivered straight to a farmer's phone.

Sample data

1. Price Forecast & Advice

Smoothes volatile market data into predictable forecast bands with simple Sell/Hold suggestions.

Advisory: HOLD (Favorable Trend)
Sample data

2. Leaf Disease Detection

Instant leaf photo analysis identifying likely diseases and actionable next steps.

Analysis: Likely Disease Identified
Sample data

3. Weather-Aware Alerts

Proactive weather notifications covering incoming rain, extreme heat, or frost risks.

Rain
Heat
Frost

Who We Serve & Hypothesized Partners

PRIMARY USERS

Smallholder Farmers

Pain Point: Unpredictable market prices and sudden crop health issues leading to avoidable financial loss.

What AgriVision Provides: Actionable sell/hold advice and instant diagnostic guidance delivered in plain language on simple web-enabled mobile devices.

HYPOTHESIZED CUSTOMERS & PARTNERS

Input Dealers, Cooperatives, NGOs & Extension Services

Pain Point: Limited real-time ground insights into regional crop diseases and market dynamics across farming communities.

What AgriVision Provides: Aggregate insights and communication channels to better serve regional smallholder farming networks.

Business model: hypotheses to validate

Our Long-term Vision

Vision

A future where every smallholder farmer can make data-informed decisions.

Our Driving Purpose

Mission

Build an honest advisor that turns public price, weather, and crop data into clear guidance.

Our Founding Principle

Honesty & Clarity

Delivering straightforward, plain-language insights without overcomplicating technology for the end user.

Measurable Objectives

All to be validated
  1. 1

    User Field Interviews

    Interview at least 3 farmers or people who work directly with farmers to ground product assumptions.

    To be validated
  2. 2

    Price Forecasting Model Evaluation

    Build a price forecast model and evaluate it against a simple baseline, such as last week's price.

    To be validated
  3. 3

    Disease Classifier Validation Metrics

    Train an image classifier on a public leaf dataset and document validation performance metrics.

    To be validated
  4. 4

    Prototype Release

    Release a working prototype targeted as a future Final Year Project (FYP).

    To be validated
Planned Offerings

Core Products & Features

Polished advisor features designed for high utility on minimal bandwidth.

Crop Price Forecast

Analyzes historical market price trends to provide plain-language sell or hold suggestions for target crops.

  • ✓ Price trend visualization
  • ✓ Sell/Hold decision aid

Leaf Disease Detection

Processes leaf photographs via computer vision models to indicate likely disease classification and immediate next steps.

  • ✓ Photo-based leaf scanning
  • ✓ Recommended treatment steps

Weather-Aware Alerts

Monitors localized environmental data to send targeted warnings regarding upcoming rain, excessive heat, or frost risks.

  • ✓ Localized risk warnings
  • ✓ Rain, heat & frost monitoring
Roadmap Strip

What's Next (Future Improvements)

Features left for future iterations beyond the initial v1 prototype.

Future Iteration

Marketplace Integration

Connecting farmers directly with verified buyers and local inputs.

Future Iteration

Digital Payments

Seamless and secure digital transactions for input procurement.

Future Iteration

Logistics Coordination

Shared transport scheduling to reduce post-harvest haulage costs.

Future Iteration

Full Offline Mode

On-device edge inference for areas with zero cellular connectivity.

Technical Architecture & Concept

The Proposed Project

Problem → Proposed Solution → Intended Benefits

Data Sources (To Be Verified)
🏛️ Government Market Price Records ☀️ Open-Meteo Public Weather API 🍃 PlantVillage Leaf Image Dataset 🏛️ Government Market Price Records ☀️ Open-Meteo Public Weather API 🍃 PlantVillage Leaf Image Dataset

Architecture Flow

Hover or tap any node to inspect system details

📊
Data Sources
Raw Input
⚙️
Pipelines
ETL Automation
🗄️
Database
Supabase / SQLite
🤖
Models
Prophet & CNNs
⚡
API
FastAPI Backend
📱
Farmer Web App
PWA on Vercel
Select or hover over any node to view architecture details.
Interactive Mockups

Concept Screens

Concept design, not a working product. All values are sample placeholder data, not real prices, diagnoses or advice.

Wheat Price Forecast Sample data
Target Crop: Wheat (Per Mound)
PKR 3,850
Forecasted Trend: Rising (+4.2%)
Recommendation: HOLD
Market conditions suggest prices may rise over the next 10 days.

Our Stack

Python (pandas, statsmodels/Prophet, scikit-learn) in Jupyter; PyTorch/TensorFlow on free Colab/Kaggle GPUs; PostgreSQL on Supabase free tier; scheduled with GitHub Actions; FastAPI on Render / Hugging Face Spaces; React PWA on Vercel; Open-Meteo weather API.

Cost: Close to zero for prototype using free tiers. (Pricing to be verified if scaled).

Progress Tracker

Company Formation Done
Problem Research & Interviews In Progress
Data Exploration Planned
Working Prototype Future FYP
No results or accuracy numbers yet.
Student Founders

Meet the Team

NED University students building AgriVision for Data Science Platform Architecture and Application.

MSA

Muhammad Sufian Adeel

CT-24257
Chief Executive Officer

Responsibilities: Vision, product direction, user interviews, coordination.

Leadership Product Strategy Python
"To lead data products that solve real problems for real farmers."
SS

Saniya Siddiqui

CT-24303
Chief Data Scientist

Responsibilities: Price forecasting and disease classifier.

Time-Series PyTorch/TF Evaluation
"To use machine learning where it makes a measurable difference."
HA

Hubba Ali

CT-24114
Chief Data Engineer

Responsibilities: Data collection, cleaning, and pipelines.

SQL pandas ETL Pipelines
"To build reliable data foundations that others can trust."
RF

Rida Fatima

CT-24060
Chief Technology Officer

Responsibilities: System design, APIs, and deployment.

Cloud Basics APIs System Design
"To design platforms that are simple, scalable and affordable."
Get in Touch

Contact AgriVision Team

Have questions about our project proposal? Send us a message.

We usually reply within a few days.
Direct Email: sufiadeel0@gmail.com
LinkedIn (Coming soon) GitHub (Coming soon)