Computer Vision News - May 2023

11 Blood Cell Classification Using CV Another approach to blood cell classification is to use data augmentation techniques. This involves applying random transformations to the input images, such as rotation, scaling, and flipping, to increase the size and diversity of the training dataset. Let's see how to implement data augmentation using the ImageDataGenerator class in Keras: from tensorflow.keras.preprocessing.image import ImageDataGenerator # Create an ImageDataGenerator with data augmentation data_gen = ImageDataGenerator(rotation_range=20, width_shift_range=0.1, height_shift_range=0.1, horizontal_flip=True, vertical_flip=True) # Load blood cell images and apply data augmentation # exercise for you! In this example, we create an ImageDataGenerator with various data augmentation options. During training, the ImageDataGenerator will randomly apply these transformations to the input images, effectively increasing the size and diversity of the dataset. We will now expand upon this example by adding data preprocessing, training, and evaluation steps. Additionally, we will provide two more coding examples: one for data augmentation using TensorFlow and another for fine-tuning a pre- trained ResNet model for blood cell classification. Image 3: Definition of white cells © Terese Winslow