9 docs tagged with "Release 2025-08"

A collaborative machine learning platform for agricultural data analysis and model training with privacy-preserving features.

This is a visualization tool for results from GNN FoodFlow Model. This portal allows you to explore the data through an interactive map, and to download the filtered results for your own research.

Harvest is a tool designed to allow multiple types of stake holders in the digital agriculture space further their own unique goals from research to increases of the bottom line. Harvest allows for the creation of pipelines where users can preprocess their data, train models on HPC resources, infer on models to get insights on farm fields, and some visualizations to give an at a glance understand of what is happening on the field.

SDK to interact with the ICICLE Model & Data Playgrounds

A project leveraging Graph Neural Networks (GNNs) to predict food flows between counties and FAF zones for economic planning, infrastructure development, and policy-making. This model predicts food trade flows between U.S. counties and Freight Analysis Framework (FAF) zones using Graph Neural Networks (GNNs). It addresses the challenges of sparsity in trade data by applying a two-stage hurdle model that distinguishes between the presence and magnitude of trade.

The Decentralized Microservice Drone System for Digital Agriculture is a distributed, scalable platform designed to orchestrate autonomous drone operations for agricultural field missions. The system captures, processes, and analyzes aerial imagery and video data to support precision agriculture, crop monitoring, and field management operations.

This project introduces an agentic approach for high-level and multi-purpose compilers

The Food Equity Access Simulation Technology (FEAST) tool — previously known as the Food Access and Strategy Simulation (FASS) tool — is a powerful platform for analyzing how changes in the food retail landscape, such as adding or removing stores, affect household food access. FEAST enables users to simulate and evaluate strategies aimed at improving food equity across communities.

This work presents a modular software pipeline and end-to-end workflow for video-based animal re-identification, which assigns consistent individual IDs by clustering multiview spatio-temporal tracks with minimal human intervention. Starting from raw video, the system detects and tracks animals, scores and selects informative left/right views, computes embeddings, clusters annotations by viewpoint, and then links clusters across time and varying perspectives using spatio-temporal continuity. Automated consistency checks resolve remaining ambiguities. Preliminary experiments demonstrate near-perfect identification accuracy with very limited manual verification. The workflow is designed to be generalizable across species. Currently, trained models support Grevy’s and Plains zebras, with plans to expand to a broader range of species.