classifying images using transfer learning
This Notebook is broken down into multiple steps:
- load the image dataset (
Oxford Flowers 102 dataset) from Tensorflow Hub and explore it, map the labels and creat a pipeline where images are resized and batched. - Build and Train an image classifier on the dataset.
- Loaded the MobileNet pre-trained network from TensorFlow Hub.
- a new untrained feed-forward network as a classifier is defined
- classifier is trained.
- loss and accuracy during training and validation are displayed
- make callbacks to save the best model
- training model is saved
- use the trained model to perform inference on flower images where we preprocess our new images to fit the model's input layer.