Robin Cole

TorchGeo: datasets, samplers, transforms, and pre-trained models for geospatial data

[ICLR 2026] RF-DETR is a real-time object detection and segmentation model architecture developed by Roboflow, SOTA on COCO, designed for fine-tuning.

Operational version of the VERCYE crop estimation library built in partnership with NASA Harvest

Geospatial-AI corn yield forecasting for the U.S. Corn Belt. Team project fine-tuning NASA/IBM Prithvi-EO-2.0-600M with LoRA, fused with weather/soil/drought features and calibrated uncertainty cones. State-level RMSE ~3–5 bu/ac — competitive with USDA WASDE. CSU Geospatial AI Hackathon 2026.