Installation

Below is the command to install with pip.

pip install -U git+https://github.com/ComputerVisionLaboratory/cvlab_toolbox

We use a Scikit-learn API so it should be pretty easy to get your code up and running. Here’s an example that should work copy&paste.

import numpy as np
from numpy.random import randint, rand
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from cvt.models import KernelMSM

dim = 100
n_class = 4
n_train, n_test = 20, 5

# input data X is *list* of vector sets (list of 2d-arrays)
X_train = [rand(randint(10, 20), dim) for i in range(n_train)]
X_test = [rand(randint(10, 20), dim) for i in range(n_test)]

# labels y is 1d-array
y_train = randint(0, n_class, n_train)
y_test = randint(0, n_class, n_test)

model = KernelMSM(n_subdims=3, sigma=0.01)
model.fit(X_train, y_train)
pred = model.predict(X_test)

print(accuracy_score(pred, y_test))