WeedGuard Ayiti

Inspiration

Haitian smallholder farmers face a critical challenge: an estimated 30-50% of crop yield loss is attributable to uncontrolled weeds, directly impacting food security and income. Additionally, inefficient manual weeding is labor-intensive and generalized herbicide application is costly and environmentally detrimental. This project was initiated to directly address these quantifiable inefficiencies by leveraging technology to provide precise, actionable solutions for sustainable agricultural development in Haiti.

What it does

WeedGuard Ayiti is an AI-powered agricultural dashboard designed for Haitian farmers, providing real-time weed detection and management strategies. It enables users to upload field imagery for AI-driven identification of weed species and their precise locations. This diagnostic capability is coupled with tailored treatment recommendations, promoting targeted intervention over broad-spectrum application. Furthermore, the platform integrates essential local weather forecasts and general agronomic advice, equipping farmers with comprehensive data for optimized decision-making and enhanced crop protection.

How we built it

The WeedGuard Ayiti system was engineered as a full-stack application, integrating a user-centric frontend developed with React.js for intuitive data interaction and visualization. The backend, powered by Flask in Python, serves as the central processing unit, managing data flow, handling image uploads, and interfacing with a specialized AI model. This AI core, meticulously trained on diverse plant imagery, is responsible for accurate weed classification. Data persistence and retrieval are managed via a robust database, while external weather information is seamlessly integrated through the OpenWeatherMap API.

Challenges we ran into

A primary challenge involved the meticulous training and fine-tuning of the AI model to achieve high accuracy in differentiating various weed species from crops under diverse field conditions, a task requiring extensive dataset curation. Furthermore, establishing seamless and secure communication protocols between the distinct frontend, backend, and AI components, particularly concerning data transfer and API key management, required some thought.

Accomplishments that we're proud of

We are particularly proud of establishing a functional AI model capable of accurate weed detection, a foundational component for precision agriculture previously less accessible to smallholder farmers. The development of an intuitive user interface, despite the complexity of the underlying technology, stands as a significant achievement, enhancing accessibility for a diverse user base. Successfully integrating disparate technologies into a cohesive, responsive system that delivers tangible value towards agricultural efficiency is a testament to our team's collaborative problem-solving and technical execution.

What we learned

Through this project, we gained invaluable insights into the practical application of artificial intelligence in real-world agricultural contexts, specifically in addressing crop management challenges. We deepened our understanding of full-stack development, mastering the intricate interplay between frontend user experience, robust backend data processing, and scalable AI integration. Furthermore, the critical importance of robust security practices, particularly in managing API keys and sensitive data through environment variables, was profoundly reinforced, shaping our approach to secure system architecture.

What's next for WeedGuard Ayiti

Our immediate future plans for WeedGuard Ayiti involve incorporating drone integration capabilities, enabling broader field surveys potentially managed by regional Department of Agriculture offices or local cooperatives. This would provide comprehensive, large-scale field assessments. We will also significantly enhance our language barrier assistance, moving beyond simple translation to provide more context-aware, culturally relevant, and verbally assisted guidance within the application. Further development will focus on expanding our weed and crop library, refining treatment efficacy algorithms, and exploring partnerships for on-the-ground farmer training and data validation.

Built With

• css
• flask
• html
• javascript
• openweathermap
• python
• python-dotenv
• react