Harvest Prediction
Use of Satellite Images to Detect Crops and Yield Prediction
Context & Objectives

Crop and yield forecasting is key for agricultural cooperatives. Knowing as early as possible the quantities harvested at a fine mesh allows them to optimize their supply chain, especially during the harvest, when tons of crops must be processed, transported, and stored in silos.

Outcome

The harvest prediction models wer used initialize and anticipate the harvest plan, which now allows for the optimization of the supply chain with a reduction of at least 10% in transport costs and CO2 emissions.

Our approach

Step 1 - Reconstruct

  • Reconstruct a unique and large database of fine-grained historical crop types and yields collaborating with agricultural cooperatives

Step 2 - Retrieve

  • Retrieve and process satellite images to build vegetation index (NDVI). These fine-grained spatial and temporal indices combined with the fine mastery of forecasting techniques allow us to make the link with the type and the yield of the crops.

Step 3 - Build

  • Build Machine Learning models to estimate the type of crop and final yield several months in advance with an error of less than 17% for more than 86% of agricultural parcels
Our experts
Yannick Léo
Partner & Data Science Director
Related Industries
No items found.
    OTHER CASE-STUDIES