Back to Projects
Farmer Vision

Farmer Vision

AIMachine LearningComputer VisionTFLite

DATE

2023-08

Overview

Farmer Vision is an Android application designed to help farmers tackle the issue of stray cattle damaging crops. The app uses a camera device that captures video footage of farmland, which is processed by a TensorFlow Lite model to detect stray animals. The real-time status of all cameras is updated and stored in a Firebase Realtime Database, allowing farmers to monitor the situation on a client device. This system helps farmers take prompt action by identifying and reporting the presence of stray cattle, minimizing crop damage and supporting their livelihoods.

Technologies Used

KotlinJavaFirebaseTensorFlow

Challenges & Solutions

CHALLENGES

  • Finding platform for release
  • Data acquisition
  • Model accuracy during deployment

SOLUTIONS

  • Android - due to its wide availability
  • Built exhaustive data scraping scripts
  • Repeated experimentation with various models

Key Learnings

  • Deploying ML models on edge devices
  • Handling realtime data
  • End-to-end solution development