Data Preprocessing for Authentication/Verification

split_into_training_gallery-probe.py

from sklearn.model_selection import train_test_split
import numpy as np 


a = ["1", "3", "4", "1", "3", "4"]
b = [[2, 4, 3],[0, 4, 5], [8, 9, 5], [5, 4, 5],[8, 4, 5], [2, 9, 5]]
y_train = np.array(a)
x_train = np.array(b)


# Step 2: Select 2 random classes for training
random_words = np.random.choice(np.unique(y_train), size=2, replace=False)
train_indices = np.isin(y_train, random_words)
x_train_random = x_train[train_indices]
y_train_random = y_train[train_indices]
#print(len(train_indices))
#print(train_indices)
#print(x_train_random)
#print(len(x_train_random))
#print(y_train_random)

# Step 3: Select the remaining 1 class for the gallery set
remaining_words = np.setdiff1d(np.unique(y_train), random_words)
gallery_indices = np.isin(y_train, remaining_words)
x_gallery_probe = x_train[gallery_indices]
y_gallery_probe = y_train[gallery_indices]
#print(x_gallery_probe)
#print(len(x_gallery_probe))
#print(y_gallery_probe)
#print(len(y_gallery_probe))


# Create a list containing both arrays
data1 = [x_train_random, y_train_random]
data2 = [x_gallery_probe, y_gallery_probe]


gallery, probe = train_test_split(
    x_gallery_probe, test_size=0.50, random_state=42, stratify=y_gallery_probe)


print(gallery)
print(probe)