Collection of plugins to integrate FiftyOne and ClearML

Plugin	Name	Desc	Has additional requirements
Clearml Export	clearml-export	Export from FiftyOne and create new datasets and dataset versions in ClearML	+
Dataset Splitter	dataset-splitter	Attach split tags (e.g. train,val) to images based on image hash. Tags can be used during export to ClearML (currently only YOLOv5 dataset format)
Zip extractor	zip-extractor	Extract images from zip to host machine to import images in FiftyOne (currently with poor optimization)
Minio importer	minio-importer	Import files from Minio	+

Plugin installation

1. Download plugin

fiftyone plugins download \
    https://github.com/HitogamiAG/fiftyone-plugins \
    --plugin-names HitogamiAG/clearml-export

2. Install the additional requirements

fiftyone plugins requirements HitogamiAG/clearml-export --install

3. Set environemnt variables

See .env_example

4. Initialize ClearML on this machine

In CLI: clearml-init and follow instructions

Development

1. Clone repository

git clone https://github.com/HitogamiAG/fiftyone-plugins && cd fiftyone-plugins

2. Create venv and install dependencies

python -m venv .venv && source .venv/bin/activate && pip install fiftyone clearml minio

3. Set environemnt variables

See .env_example

4. Set FIFTYONE_PLUGINS_DIR to plugins folder in repo

export FIFTYONE_PLUGINS_DIR=$(pwd)/plugins

5. Run FiftyOne

python tests/fiftyone_local.py

There is no need to restart FiftyOne to apply changes in plugins' source code. Just call plugin function again.