Smallholder Agricultural Landscape Understanding (Papers Track)
Radhika Dua (New York University); Aditi Agarwal (Google DeepMind); Alex Wilson (Google Research); Hoang Tran (Google); Nikita Saxena (Google DeepMind); Ishan Deshpande (Google DeepMind); Bogdan Floristean (Google); Neelabh Goyal (Google); Ramya Cheruvu (Google); Ujwal Singh (Google); Jitendra Jalwaniya (Google); Amandeep Kaur (Arizoana State University); Batchu Venkat Vishal (Google Research); Yan Mayster (Google); Gaurav Aggarwal (Jio Platforms Limited); Alok Talekar (Google DeepMind); Vaibhav Rajan (Google DeepMind)
Abstract
Comprehensive agricultural landscape understanding is critical for addressing global challenges in food security and climate change. This requires mapping not just crop fields, but also vital features like trees and water bodies which form an intricate mosaic in smallholder systems dominating the Global South. Previous efforts to develop such land use maps have been limited by a narrow focus on methods for field delineation only, and also do not develop robust post-processing steps essential for real-world deployment. This work addresses these limitations by presenting the first national-scale agricultural mapping system that moves beyond simple field delineation to provide comprehensive, multiclass segmentation of fields, trees, and water bodies, which are publicly accessible, enabling a wide range of applications and advancing global sustainability development goals.