driver.py

import cv2
import kmeans as km
import videoprocessing as vp

def main():
    #Video processing
    path = 'resources\\clip3.mp4'
    target_resolution = (100,150)
    target_fps = 2

    print('Processing video...')
    video_array = vp.video_to_array(path,target_resolution,target_fps)
    print('Done.')

    #K-means algorithm
    X = video_array
    k = 5
    max_iter = 10

    print('Performing K-means...')
    centroids = km.initialize_centroids(X, k)
    for i in range(max_iter):
        print('iteration '+str(i+1)+' out of '+str(max_iter))
        closest_centroids = km.find_closest_centroids(X, centroids)
        centroids = km.update_centroids(X, k, closest_centroids)
    print('Done.')

    #display centroids
    for i in range(k):
        image = centroids[i,:].reshape(target_resolution[0],target_resolution[1],3)
        #cv2.imshow('centroid '+str(i), image)

    #display thumbnails
    closest_points = km.find_closest_points(X, centroids)
    for i in range(k):
        image = X[closest_points[i],:].reshape(target_resolution[0],target_resolution[1],3)
        cv2.imshow('thumbnail '+str(i), image)

if __name__ == '__main__':
    main()