SkyHarvest

Inspiration

Agriculture remains the backbone of India's economy, yet farmers face critical gaps in real time, accessible information: crop health, irrigation timing, market prices, and government schemes. Existing platforms are fragmented, often inaccessible to those with low digital literacy, and fail to provide personalized, actionable insights. The idea for SkyHarvest emerged during the hackathon as we asked a simple but powerful question: "What if the sky itself could guide every farmer's field?" By combining satellite imagery, Al-driven crop identification, and multilingual voice-first assistance ,we envisioned a platform that not only empowers farmers with macro-level insights for smarter agricultural planning.

What it does

SkyHarvest has been created with the farmer in mind, focusing on practical solutions that tackle the most urgent challenges faced in rural India. The current prototype offers the following main features:

Satellite AI Crop Identification: Using freely available satellite datasets like Sentinel-2, SkyHarvest employs machine learning models to identify the crop growing on each farmer’s plot. This eliminates the need for manual surveys or reliance on middlemen, giving farmers quick and dependable crop recognition with just a tap. Farmers receive automated insights that confirm the crop type, which builds their trust in the platform while setting the stage for customized advisory services.
Personalized Advisory Feed: Based on the identified crop, SkyHarvest creates a personalized stream of advisories that directly address the farmer’s current needs. These include:
Irrigation schedules that reduce water waste and improve yield.
Weather alerts that help farmers prepare for rainfall, heat waves, or frost.
Mandi (market) price updates for local trading centers, giving farmers real-time pricing information to make better selling choices.
Information on government schemes relevant to the farmer’s crop and region, helping to cut out middlemen for updates. The advisory feed is simple, clear, and actionable, ensuring farmers receive not only data but also straightforward next steps.
Low Tech Accessibility Layer: Understanding that many rural farmers still use feature phones and low-bandwidth networks, SkyHarvest was designed for inclusivity. Farmers can currently receive updates through:
SMS alerts for important information like irrigation reminders or mandi prices.
A lightweight mobile app interface (Android-first), which functions well on basic smartphones and in areas with weak connectivity.

How we built it

Challenges we ran into

Data Resolution & Accuracy: Finding the sweet spot between satellite image resolution (10m–30m) and field-level accuracy.
Multilingual Diversity: Making AI voice responses work in multiple regional languages with small training set.
Connectivity challenges - Balancing offline-first with real-time advisory relevance. Time pressure building a working prototype within hackathon time frame, especially making sure we had a full end to end integration.

Accomplishments that we're proud of

Successfully showed how to identify crops using satellite data from real Sentinel-2 datasets for Indian farmland. Launched a working advisory prototype that is available through SMS and an app. Created a multilingual voice assistant that can handle crop-related questions in Hindi and English. Received positive feedback from mentors for addressing grassroots usability.

What we learned

Enhanced technical competence in Geospatial AI, Crop analytics, and Rural information systems. How to be inclusive with AI/ML model’s with low tech interfaces (SMS/voice) gained! First, we recognized how critically important it is to design for humans — working with tech that fits into farmer realities, not the other way around. Unlocked the power of linking high-tech satellite imagery with mass rural empowerment.

What's next for SkyHarvest

Accuracy Improvement: Use higher resolution imagery from Planet Labs and ISRO’s Bhuvan. Refine models with ground-truth data from farmer cooperatives.

Expanded Multilingual Support: Train the voice assistant in over 10 Indian languages and dialects.

Predictive Analytics: Detect crop diseases, pest outbreaks, and water stress early. Use AI for yield forecasting at village and district levels.

Integration with Government and NGOs: Work together for real time scheme delivery, including subsidies and crop insurance. Offer policymakers district-level dashboards for decision making.

Sustainability and Scale: Generate revenue through partnerships with banks, agri-input firms, and state departments. Provide open APIs for ecosystem collaboration.