DataFrames.

A DataFrame is a tabular structure used to organize records in rows and columns. It differs from the picture displayed at one moment in a video, but the two can work together: a row can store that picture's timestamp, dimensions, file reference, and analysis results. Keep the table's schema separate from the image representation. Storing a frame identifier is different from storing its pixels, and a missing analysis value is different from a valid result that found nothing.

This archive focuses on building useful frame datasets without losing the original media relationships. Examine how identifiers, time units, output paths, and transformation history are represented before joining or aggregating records. The DataFrame and Python workflow overview connects those choices. To check a proposed schema, select a few rows and trace them back to the source video and exported images. If that route is unclear, improve the records before adding more calculated columns or model outputs.