Out Of The Box pipelines for LOST image annotation tool
This repo is currently under construction!
Request single image annotations for all images of a specified data source.
Request multi image annotations for all images of a specified data source.
Request all annotations from a dataset file (csv or parquet) in lost_dataset format again.
Request all annotations from a dataset file (csv or parquet) in lost_dataset format again.
This pipeline represents a two stage annotation process. In a first stage bbox annotations are collected and in the second stage this bboxes will be labeled by MIA.
In first stage use MIA to sort out images that are not of interest (e.g. empty images). In second stage annotate the selected images from stage one with SIA.