The DS@GT ARC team submitted an ensemble of U-Net and the Prithvi-2.0 geospatial foundation model to predict viticulture potential in Southern France for the ImageCLEF AI4Agri 2026 competition.

  • The submission targeted Subtask 1 of the ImageCLEF AI4Agri 2026 challenge.
  • The model achieved a ±1 accuracy score of 68.32 on the leaderboard.
  • This result ranked the team 2nd among 7 participating teams.
  • The implementation is publicly.