3 new models designed to detect field boundaries for smallholder farmers available on Radiant MLHub
Louisa Nakanuku-Diggs
I am excited to announce the winners of the NASA Harvest Field Boundary Detection Challenge! Out of 730 participants, the top 3 machine learning models have been recognized for their exceptional performance in detecting field boundaries for smallholder farmers.
You can now find these winning solutions on Radiant MLHub. Check out the details and links to each model below👇🏽
🥇Spatio-Temporal Attention-based Unet for Field Boundary Detection. This solution is a single 10-fold modified Regnetv-Unet developed in Pytorch. Created by Muhamed Tuo and Azer Ksouri.
🥈 Harvest Ensemble Segmentation Model for Fields. It's based on a number of pre-trained models & decoders, trained with full data without validation using sam optimizer with adamw as the base optimizer to limit overfitting. Created by Bojesomo Alabi.
🥉 Borderline: A segmentation model for fields. This solution was built using torch. Created by Hoang Truong, Tien Dung & MG Ferreira.