This directory contains scripts for creating new bounding box annotation tasks and adding annotations to the database.
We send all images that need to have their bounding boxes annotated in a flat folder zipped together.
A database in the MegaDB format should be created with any known labels and information prior to sending the images to the annotators.
The images to be annotated should be named:
dataset<dataset>.seq<seq_id>.frame<frame_num>.jpg
The annotators will only use the seq<seq_id> field to group sequences together.
If the dataset has no explicit sequence information, it might be best to have the seq_id set to a short version of the image name (with dummy_ prepended) so if the sequence information is in the file names, the sequence can appear together.