Start Date: March 8 End Date:  2 June Competition URL: Kaggle

Context

The HyperLeaf2024 challenge consists of 2410 hyperspectral images of wheat flag leaves, encompassing multiple strains and fertilizer levels and taken at different exposures and under different sunlight conditions. The challenge is aimed at developing models for fine-grained prediction of wheat field properties, plant health metrics, and yield.

Although there already exist publicly available drone and satellite hyperspectral imaging datasets. These typically have the goal of per-pixel image classification (one target per pixel). However, the power of hyperspectral imaging in full image classification and regression (one target per image) is less studied. We hope to promote this use case of hyperspectral imaging in wheat agriculture through this dataset and challenge.