With TensorFlow 2.3, Amazon SageMaker Python SDK 2.5.x and Custom SageMaker Training & Serving Docker Containers
This project shows step-by-step guide on how to build a real-world flower classifier of 102 flower types using TensorFlow, Amazon SageMaker, Docker and Python in a Jupyter Notebook. It has been tested with the Python packages in the requirements.txt on Python 3.8.5.
Clone this project from GitHub. Create a new Python virtual environment targeting Python 3.6 and above.
Install the required Python packages from the requirements.txt or install them from running the project's Jupyter notebook. Start and run the notebook with Jupyter Lab.
Note that the external flower images used in the notebook are not provided as part of the project. You could use any other free flower images at your own discretion for the evaluation of the project's flower classification model that you are going to build and deploy.
Pull requests, suggestions and feedback are welcome. For major changes or issues, please open an issue to discuss.