Primary competition visual

GeoAI Aquaculture Pond Identification Challenge by FAO and ITU

1000 CHF
2 months left
Data Analysis
Classification
Feature Engineering
GIS
487 joined
143 active
Starti
Jun 08, 26
Enrolments closei
Aug 07, 26
Closei
Aug 16, 26
Reveali
Aug 16, 26
About

You are provided with a tabular dataset consisting of 953 entries for training and 858 entries for testing respectively. The tabular features are derived from freely accessible satellite data across two pilot regions. They are organised as 12 monthly composites for each location, specified by latitude and longitude. The features include:

  • Sentinel-1 (SAR) - radar backscatter bands (VH, VV) capturing surface texture and moisture
  • Sentinel-2 (Optical) - multispectral surface reflectance bands capturing the spectral signature of water bodies and surrounding vegetation

The target variable is binary: aquaculture pond (1) or other (0).

The latitude and longitude should only be used to download data. They SHOULD NOT BE USE AS FEATURES.

Files
Description
Files
Shows the structure of the sample submission file
Labelled satellite observations for 963 locations used to train the model.
AI Trustworthiness Evaluation document
Unlabelled satellite observations for 858 locations used to evaluate the model.