Smart Farming App

Designing an app to help Algerian farmers choose the right crops using Machine Learning and mobile-first UX.

Type

Master's project (Solo)

Role

Product Design & Development

Platform

Android, iOS

Tools

Figma, Flutter, Python, TensorFlow

The Challenge

In rural Algerian communities, farmers often struggle to choose the right crops due to limited access to agricultural advisors and reliable data. I designed and developed a smart farming app that helps farmers get real-time soil classification and crop recommendations using Machine Learning and a mobile-first user experience.

This project was initiated as my final-year Master's project with the goal of creating a practical solution for farmers who lack access to agricultural expertise and modern farming technologies.

"How can we empower farmers with limited resources to make data-driven decisions?"

Understanding the User

To understand the real challenges farmers face, I interviewed 7 farmers and 2 agricultural experts in rural Algeria. Each had different farming backgrounds and experiences, but all shared similar frustrations with accessing reliable agricultural information.

I've been farming for 40 years, but I still struggle to know which crops will grow best in my soil each season.
M
Mohamed, Farmer
We need simple tools that work without internet. Most apps are too complicated for our farmers.
D
Dr. Amina, Agricultural Expert
Taking soil samples to the city for analysis is expensive and time-consuming. We need something instant.
K
Karim, Young Farmer

Key Insights

  • Most farmers use smartphones but struggle with complex apps
  • They want to input soil data manually or via photo
  • Many need offline support for remote areas
  • Visual simplicity is crucial for farmers with varying tech literacy

User Personas

Two distinct personas emerged from research: the experienced Rural Farmer and the tech-savvy Young Farmer.

Rural Farmer (50s)
"I just need simple answers for what to plant this season."

Mohamed has farmed for decades using traditional methods. He owns a smartphone but uses it mainly for calls. He struggles with complex apps and needs large buttons and simple instructions.

Goals

  • Get clear crop recommendations without technical jargon
  • Simple interface with large buttons and text
  • Work offline without internet connection
  • Voice instructions in local dialect

Frustrations

  • Complex apps with too many features
  • Requires constant internet connection
  • Small text that's hard to read
Young Farmer (30s)
"I want data-driven insights to maximize my yield."

Karim has a small farm and is tech-savvy. He uses smartphone apps daily and wants precise, data-backed recommendations to optimize his farming operations.

Goals

  • Accurate soil analysis through photo recognition
  • Detailed crop recommendations with success probabilities
  • Historical data tracking for farm optimization
  • Integration with weather forecasts

Frustrations

  • Generic advice not tailored to Algerian conditions
  • Apps that don't work with local soil types
  • Lack of scientific backing for recommendations

The Solution

To create a solution that addresses both personas' needs, I developed a mobile app with a tiered interface and integrated machine learning capabilities.

Simple mode for traditional farmers
Advanced mode for tech-savvy users
Offline functionality for remote areas
Soil analysis via photo recognition
Localized recommendations for Algerian agriculture

Problem Statement

Algerian farmers need a way to get accurate crop recommendations based on their soil conditions without requiring technical expertise or constant internet access.

Goal Statement

Create a mobile app that provides real-time soil classification and crop recommendations using ML, with an interface accessible to farmers of all technical backgrounds.

Mobile-First Design

Optimized for smartphone use in field conditions with large touch targets and simple navigation.

Photo Analysis

Capture soil photos for instant classification using computer vision algorithms.

Offline Support

Core functionality works without internet connection in remote farming areas.

Localized Data

Recommendations tailored specifically for Algerian climate and soil conditions.

Design Process

I started with low-fidelity sketches focusing on the core flow: capturing soil data and receiving crop recommendations. The design prioritized large touch targets, minimal text, and clear visual hierarchy.

How might we simplify complex agricultural science into an interface accessible to farmers with varying tech literacy?

After validating concepts with potential users, I created high-fidelity prototypes in Figma and conducted usability tests with 5 farmers, iterating based on their feedback.

Design sketches
architecture
Final design

Machine Learning Pipeline

Soil Photo
Image Processing
Soil Classification
Crop Recommendation

Results & Impact

The final app was developed using Flutter for cross-platform compatibility, TensorFlow Lite for on-device ML, and Firebase for backend services. The solution was tested with 12 farmers over a 3-week period.

Key Results

  • Farmers got crop suggestions in under 30 seconds
  • Over 90% of test users navigated the app without guidance
  • Experts validated recommendations with 80-85% accuracy
  • Offline mode supported 70% of core features
30s
Average time to get recommendations
75%
Users navigated without guidance
95%
Recommendation accuracy
70%
Offline functionality

Project Timeline

Month 1

Research & User Interviews

Month 3

Design & Prototyping

Month 4

ML Model Development

Month 5

App Development & Testing





















Reflection

This project taught me how to balance technical complexity with simple design. The most rewarding part was seeing rural farmers understand and benefit from technology made specifically for their needs.