readme.md

Image Segmentation

A common data source and loader will be used in this tutorial between PyTorch and TensorFlow. Run the data.py first to download the data from the Oxford-IIIT Pet Dataset.

conda activate comp0197-tf  # or comp0197-pt
python data.py

Once download completes, in the Python console, example images and segmentation labels can be saved for visualisation.

import numpy as np
from PIL import Image
from loader import H5ImageLoader

DATA_PATH = './data'

images,labels = next(iter(H5ImageLoader(DATA_PATH+'/images_train.h5', 10, DATA_PATH+'/labels_train.h5')))
image_montage = Image.fromarray(np.concatenate([images[i] for i in range(len(images))],axis=1))
image_montage.save("train_images.jpg")
label_montage = Image.fromarray(np.concatenate([labels[i] for i in range(len(labels))],axis=1))
label_montage.save("train_labels.jpg")