Welcome to the AgML Community

[masonearles]

Agricultural machine learning will not advance through model scale alone. It also needs discoverable and interoperable data, clear metadata and label ontologies, quality diagnostics, comparable benchmarks, domain-aware evaluation, and collaboration across agricultural and technical communities.

This is a public space for building that infrastructure together.

Who belongs here?

We welcome:

• Agronomists, plant scientists, growers, and extension professionals
• Dataset creators, curators, and data stewards
• Machine learning, computer vision, foundation-model, and VLM researchers
• Engineers, students, educators, and open-source contributors
• People working in adjacent fields such as geospatial AI, remote sensing, biodiversity informatics, and biomedical imaging

You do not need to write code to participate. A field use case, metadata question, dataset, critical evaluation, or clearly explained problem can be an important contribution.

What are we working on?

Initial priorities include:

• Discovering and standardizing agricultural datasets
• Improving metadata, provenance, licensing, and interoperability
• Assessing dataset quality, diversity, leakage, and annotation reliability
• Benchmarking foundation models and vision-language models
• Expanding evaluation across classification, detection, segmentation, counting, and reasoning
• Connecting imagery with sensor, soil, weather, management, harvest, spatial, and temporal data
• Measuring geographic and real-world generalization

How to participate

1. Introduce yourself and share what you are working on.
2. Join a discussion where your experience is relevant.
3. Share a dataset, schema, benchmark, use case, question, or proposed experiment.
4. Help turn a mature discussion into documentation, an issue, a dataset improvement, a benchmark run, research, or code.

Start with the community README and contribution guide.

How this space will operate

Discussions are for questions, collaboration, and early proposals. Once an idea becomes actionable, maintainers will help summarize it and move the work to the appropriate AgML repository or Hugging Face resource. Outcomes will be shared publicly, and contributors will be credited.

Please follow the AgML Community Code of Conduct.

We are building this community in public. Suggestions for improving the process itself are welcome.