The purpose of this project is to demonstrate various use cases for transfer learning in object classification, object localization and object recognition. Popular models are used on a bunch of datasets of varying sizes. The tooling is centered on PyTorch for model implementation and MLFlow for experiment tracking.
TBD
The project is centered around a package of utility functions installable as cvision. Recommended approach is to use an editable install in a running virtual environment.
- Activate your venv.
- Install with
pip install --editable .from project root.
MLFlow tracking server is used for centralized logging for all model training runs. The server is packaged as a Flask app that can be started directly on localhost or via Docker.
To use a bare metal instance:
- MLFlow package is installed along with other project dependencies.
- Run server with
mlflow server --host 127.0.0.1 --port 8888.
To use Docker:
TBD