Clustering in Python- Image Clustering

Question

I want to cluster the images using K Means or other algorithm(suggestion required).

The problem is like this- I want to cluster images into 3 clusters (nature, sunset, water). I loaded all the images using os.listdir() and then converted all of the images into arrays (RGB) and then created a data frame which contains three columns - ID, Image_array, Label.

Now, when I use K Means clustering, providing n_clusters = 3, it shows this error:

from sklearn.cluster import KMeans kmeans = KMeans(n_clusters = 3).fit(img_array) ERROR = Found array with dim 4. Estimator expected <= 2.

Now, I need your help in this clustering problem. The data frame that I created looks like this

img_array = []

path = "C://Users/shivam/Desktop/freelancer/p22/data/green_nature/"
for f in os.listdir('.'):
    if f.endswith('.jpg'):
        img = Image.open(f)
        data = np.asarray(img, dtype='uint8')
        img_array.append(data)


df = pd.DataFrame({'image_arrays':img_array})
df['id'] = range(1, len(df) + 1)

Alianse777 Alianse777 · Accepted Answer · 2017-04-03T13:57:03

This happen because you pass 4-dim array while 2-dim expected. 'img_array.shape' should be like this (n_samples, n_features). You need to use feature extraction algorithm.

This can be done via scikit-image module. You need to convert images to greyscale format. Code:

import skimage.feature as feature
img_converted = []
for i in range(len(img_array)):
    img_converted.append(feature.hog(img_array[i]))
model.fit(np.array(img_converted))

Documentation: http://scikit-image.org/docs/dev/api/skimage.feature.html#hog

Clustering in Python- Image Clustering

2 Answers