KT
PORTFOLIO INDEX
KRISH TRIVEDI · AHMEDABAD, GUJARAT, INDIA
Back to All Projects
SMART INDIA HACKATHON 2025 NOMINEE APPLIED COMPUTER VISION & LLM

KisanMitra – AI Krishi Officer

Comprehensive AI agricultural assistant developed for Smart India Hackathon 2025. Delivers instant crop disease identification from uploaded plant images, multilingual voice advisories in local dialects, and real-time agricultural guidance.

kisanmitra.ai-krishi.sih2025.in/diagnose
SIH 2025 Evaluation
KisanMitra AI Krishi Officer Architecture

Problem Statement

In rural Indian agriculture, farmers lose an estimated 20–35% of crop yields annually to plant diseases and pest infestations. Traditional extension services cannot reach remote farmers quickly, and language barriers make text-based agronomy apps unhelpful for non-literate cultivators.

Project Objective

Create an intuitive, zero-friction AI Krishi Officer that allows farmers to take a quick photo of an infected leaf, receive an instant pathology diagnosis with 90%+ confidence, and listen to curative instructions spoken in their mother tongue (Gujarati and Hindi).

Solution Overview

The KisanMitra system integrates two core AI modules orchestrated via a high-performance Python Flask backend:

1. Computer Vision Diagnostic Engine

Trained on extensive crop pathology datasets covering wheat, tomato, cotton, and paddy diseases. Detects leaf spot, rust, blight, and powdery mildew with bounded region of interest highlight.

2. Multilingual Advisory & Audio Synthesis

Transforms diagnostic output and weather forecasts into clear, dialect-specific voice advisories. Farmers can ask follow-up questions using voice queries without needing to type.

Key Architectural Features

Instant Image Upload Diagnostics: Mobile camera optimization and sub-second foliage pathology classification.
Multilingual Voice Enabled: Automatic Speech Recognition (ASR) and Text-To-Speech (TTS) in Gujarati and Hindi.
Dosage & Treatment Calculation: Provides exact fungicide/pesticide dosage per liter of water based on acreage.
SQLite History & Outbreak Tracking: Logs historical diagnoses to map regional agricultural pest outbreaks.

Verified Technology Stack

Python Flask API PyTorch / ResNet OpenAI API Computer Vision Multilingual Speech TTS SQLite Pillow (PIL) Gunicorn

My Engineering Contribution

  • • Lead AI Project Developer for the Smart India Hackathon team.
  • • Built the Python Flask API handling image ingest and inference pipelines.
  • • Integrated the OpenAI prompt synthesizer for localized dosage instructions.
  • • Designed the Gujarati and Hindi audio voice response pipeline.
  • • Presented live system demonstrations during hackathon evaluations.

Hackathon Recognition

Competition: Smart India Hackathon 2025
Category: Agriculture & Rural Development
Status: Nominee Finalist
← White Swan Holidays View Full Projects Catalog