In the current workflow, parquets are used to save output data structures containing measurements from detected bubbles in images. However, these parquets are used to save data objects in several places (e.g. withbubblesam bounding boxes and contour arrays). These values should be stored in a more idiomatic way (i.e. as separate columns of a dataframe) to simplify downstream dataset handling.
In the current workflow,
parquetsare used to save output data structures containing measurements from detected bubbles in images. However, these parquets are used to save data objects in several places (e.g. withbubblesambounding boxes and contour arrays). These values should be stored in a more idiomatic way (i.e. as separate columns of a dataframe) to simplify downstream dataset handling.