visualize.py

import matplotlib.pyplot as plt
import glob
import os
import numpy as np
# level80_case0035_image


base_path = os.path.join("output_original_single_channel", "predictions-ood")
storage_path = os.path.join("output_original_single_channel", "predictions-ood-processed")
if not os.path.exists(storage_path):
    os.mkdir(storage_path)
for level in range(48, 90):
    for sigma in range(1, 11):
        images = []
        for filename in glob.glob(os.path.join(base_path, f"level{level}_case0035_{sigma}_*_prediction.npy")):
            # im = plt.imread(filename)
            im = np.load(filename)
            im[im != 0] = 1
            images.append(im)
        preds = np.stack(images)
        preds.shape
        variance = np.var(preds, axis=0)
        # variance = (variance - np.min(variance)) / (np.max(variance) - np.min(variance))
        print(f"{level} {sigma}")
        plt.imshow(variance, cmap="magma")
        plt.imsave(os.path.join(storage_path, os.path.splitext(os.path.basename(filename))[0].replace("prediction", "uncertainty") + ".jpg"), variance, cmap="magma")
        plt.show()