Explain why geometry is needed
A model using images alone may need different metadata from a project converting pixels into spatial quantities or comparing multiple cameras. State the intended operation before requesting calibration. This helps distinguish essential measurements from useful background. A dataset should not be described as geometrically suitable merely because it includes several camera streams or high-resolution images.
Identify the relevant configuration
Request camera identity, placement, image dimensions, and applicable intrinsic or extrinsic information where needed. Ask which episodes each configuration covers and whether cameras or tools were moved. A calibration file separated from its setup history can be difficult to apply correctly. Preserve configuration identifiers at the episode level rather than attaching one unexplained file to an entire collection.
Track image transformations
Cropping, resizing, rotation, and other processing can change how supplied geometry relates to delivered pixels. Document the actual operations and identify whether metadata has been updated accordingly. If original and processed streams are included, keep their relationship explicit. Spatial labels should also reference the correct image version, especially when derived files are created after annotation.
Review evidence and limitations
Ask how the information was obtained and what checks the source performed. Inspect a representative configuration through the intended pipeline and record unresolved questions. Available calibration metadata is not a guarantee of suitability for every spatial task. Include required geometry, permitted uncertainty, and configuration coverage in the inquiry so source fit can be assessed before a collection is selected.
Review checklist
- Connect geometry requirements to a task
- Link calibration to configurations
- Record image transformations
- Review source evidence and limitations