https://github.com/sproc01/starthack_bibo
Web app developed for the start hack 2025
https://github.com/sproc01/starthack_bibo
ai backend frontend python3 react webapp
Last synced: about 2 months ago
JSON representation
Web app developed for the start hack 2025
- Host: GitHub
- URL: https://github.com/sproc01/starthack_bibo
- Owner: Sproc01
- Created: 2025-03-19T20:02:44.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-03-23T11:58:14.000Z (over 1 year ago)
- Last Synced: 2025-03-23T12:29:13.364Z (over 1 year ago)
- Topics: ai, backend, frontend, python3, react, webapp
- Language: Python
- Homepage:
- Size: 27.2 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# START HACK 2025 - BIBO X Syngenta Challenge
## 🌱 Project Overview
This is an AI-enabled solution developed during START HACK 2025 in St. Gallen for Syngenta's challenge: "Nature helps nature: Use AI to improve global farming through nature-powered innovation." Our platform helps farmers in India and Brazil make more informed decisions about biological product usage based on their specific conditions.
## 💡 Problem Statement
Farmers worldwide face a critical challenge in selecting optimal crop treatments due to:
- Limited data on treatment efficacy across varying climates and soil conditions
- Difficulty predicting how treatments perform in specific local environments
- Inability to accurately assess which solutions will maximize yields while minimizing costs
This knowledge gap leads to suboptimal yields, wasted resources, and reduced farm profitability.
## 🚀 Solution
Our project addresses these challenges through an AI-powered platform.
The key components of our solution are:
### 1. Comprehensive Historical Data Collection
- Downloaded and processed weather forecast data spanning the last 50 years
- Included parameters such as temperature, precipitation, humidity, and soil features
- Organized data by geographic regions to ensure localized predictions
- Created a robust dataset that captures seasonal patterns and climate change trends
### 2. Advanced AI Prediction Model
- Implemented a Mixture of Experts (MoE) model architecture that specializes in different risk factors
- Trained the model to predict potential risks for various crop types based on weather patterns
- Achieved up to 99% prediction accuracy for major risk events up to 3 weeks in advance
- Lightweight models allow for real-time predictions and updates + low computational costs and environmental impact helped by heavy caching
### 3. Accessible Farmer Interface
- Developed a user-friendly web application for farmers to access predictions
- Implemented location-based risk assessments for specific fields and crops
- Created visualization tools to communicate risk levels through intuitive charts and maps
- Provided recommendations for optimal biological product usage based on predicted risks
- A powerful funnel which converts the tool in customer fidelization and increased sales for Syngenta
### 🖼️ Screenshots
## 👥 Team
- [Lorenzo Asquini](https://github.com/lorenzo-asquini)
- [Alberto Pasqualetto](https://github.com/albertopasqualetto)
- [Michele Sprocatti](https://github.com/Sproc01)
- [Riccardo Zuech](https://github.com/ZuechR)