The aim of this README is to give you alle necessary information on how to use the provided data and thus enable you to reproduce our results.
To use this repository please first visit Facebook AI Research's Detectron repository to download and install the used framework.
- write a manual on how to use the provided data
- upload config files necessary to reproduce training strategies
- amend output dirs in config files to match future directory structure
- upload evaluation scripts to make sure evaluation is done the same way
- upload all files changed inside detectron such as dataset_catalog.py and task_evaluation.py
- upload the annotation files
- create new Dockerfile
- test new Dockerfile