What agricultural data and metadata should AgML support next?

[masonearles]

AgML has made hundreds of public agricultural vision datasets easier to discover and use. The next challenge is not simply adding more datasets. It is making agricultural data more interoperable, interpretable, and useful across crops, regions, sensors, seasons, tasks, and research groups.

We would like the community's help defining what AgML should support next.

Questions for the community

• Which crops, production systems, geographies, environments, and agricultural tasks are poorly represented?
• Which metadata fields are essential for interpreting or reusing an agricultural dataset?
• How should we represent geography, collection period, phenology, sensor, platform, imaging conditions, management history, and annotation provenance?
• How should imagery connect with soil, weather, environmental sensors, irrigation, management, harvest, yield, genomic, or other data?
• Which existing agricultural standards or ontologies should AgML align with rather than reinvent?
• What makes a data pipeline genuinely reusable across institutions and projects?
• Which licensing, privacy, sovereignty, or attribution constraints need to be modeled explicitly?

Share a concrete use case

If possible, describe:

• The agricultural problem and intended user
• Crop, system, and geographic context
• Data modalities and approximate scale
• Current file formats or schema
• Temporal and spatial resolution
• Important calibration, units, missingness, and provenance information
• How different observations should be linked
• Whether the data or a representative schema can be shared publicly
• The decision, model, or analysis the pipeline should support

A public data dictionary or small de-identified example schema can be more useful initially than uploading a full dataset.

We will synthesize recurring needs from this discussion into candidate metadata extensions, documentation, and focused implementation or research proposals.