Crop Yield Prediction

Repository contains a solution for CGIAR Crop Yield Prediction Challenge.

Models utilize satellite multi-spectral images, climate and soil data for fields and predict crop yield in tons per acre.
There are 2 models that differ in sequential data processing approach:

1. CropNet - processes sequential data with LSTM.
2. CropNet2 - processes sequential data with CNN.

Scematic models arcitecture is repsesented in the figure below:

Examples:

• Dataloader usage examples
• Model training examples

Data

• Data for training is provided on a copmetition page (data).