Benchmark Image Dataset Collection And Preprocessing

• MNIST dataset and its variants - 12000 train, 50000 test

  • MB: MNIST basic dataset
  • MBI: MNIST background image - A patch from a black and white image was used as the background for the digit image
  • MDRBI: MNIST digits with rotation and background image - The perturbations used in MRD and MBI were combined.
  • MRB: MNIST random background - A random background was inserted in the digit image
  • MRD: MNIST rotated digits - The digits were rotated by an angle generated uniformly between 0 and 360 radians.

• CONVEX dataset - 8000 train, 50000 test

Download the datasets and put the files under the root directory of your project as shown in the following picture.

# import the loader tool
from bidcap.utils.loader import ImagesetLoader
# import mb. Pass the dataset name described above as the first parameter
data = ImagesetLoader.load('mb')
# training images
data.train['images']
# training labels
data.train['labels']
# test images
data.test['images']
# test labels
data.test['labels']