FarmLens

Problem

Smallholder farmers spot crop disease late, when treatment costs the most and yield is already lost.

Architecture

Flutter/mobile and web clients → FastAPI inference service → SQL records for predictions. Image preprocessing and model inference run server-side with structured response payloads.

Outcome

Working end-to-end pipeline: upload, classify, score confidence, and persist the result to a queryable prediction history.

Engineering focus

• —Balancing model accuracy with inference time on consumer hardware
• —Designing a clear upload → result flow for non-technical users
• —Keeping prediction history consistent across API and database layers

What it does

• —Leaf image classification with confidence scores
• —Structured prediction records and API integration
• —Real-time image processing pipeline
• —Cloud-ready FastAPI service architecture

Built with

• Python
• FastAPI
• Flutter
• Node.js
• SQL
• Computer Vision
• ML