This hackathon challenges participants to build a crop health monitoring and yield estimation pipeline using Capella Space's X-band Synthetic Aperture Radar (SAR) imagery. The goal is to design a novel crop health index and estimate relative yield potential for farms in Vadodara, using multi-temporal Capella X-band SAR imagery captured across four dates through October, alongside village and farm boundary shapefiles.
Crop health monitoring today relies heavily on optical satellite imagery, which fails during cloud cover — a major limitation during India's monsoon-affected Kharif season. Synthetic Aperture Radar (SAR) sees through clouds, but X-band SAR specifically remains largely unexplored for agricultural health and yield applications. This hackathon asks you to build a crop health index and a relative yield estimate for individual farm plots using Capella Space X-band SAR imagery — with no prescribed formula, no labeled ground truth, and no fixed methodology.
What You're Given 4 temporal Capella X-band SAR scenes (HH polarization), captured on June 6, 2025, June 19, 2025, August 14, 2025, and October 13, 2025. Data includes farm boundary shapefiles (966 farms) across 1 village near Vadodara and village boundary shapefiles. Crops present include Rice, Cotton, Maize, Bajra, and Groundnut. Participants are expected to carry forward the crop classification map from Round 1 to determine the dominant crop per field. Auxiliary data (weather, optical satellite imagery, DEM, soil) may be brought in, but Capella SAR imagery must remain the primary dataset.
Submission Requirements
A valid submission must contain a Kaggle Writeup (including title, subtitle, detailed analysis, and media gallery), an attached public notebook, and an attached project link. The notebook must include the full pipeline and produce a final output CSV with columns: village_id, farm_id, crop_type, health_index, and yield_estimate_to_date. The Writeup should not exceed 4 pages.
Tracks and Awards Remote Sensing: Crop Health & Yield Estimate farm-level crop health and relative yield potential using multi-temporal Capella X-band SAR imagery. Submissions are judged by a panel against a published Evaluation Rubric, not scored on a leaderboard. Top-performing teams advance to Phase 3: the Grand Finale in Goa, September 2–3, 2026. Prize details are available on the Hackathon Website.
Evaluation The rubric (100 points total) assesses Technical Soundness (25 points), Methodological Creativity (20 points), Validity & Plausibility of Results (20 points), Village-Level Aggregation & Coverage (10 points), Documentation & Reproducibility (15 points), Presentation Quality (10 points), and Required Elements.