Kmeans-Clustering

K-means clustering is a popular unsupervised machine learning algorithm used for grouping similar data points into clusters. It aims to partition a dataset into K clusters, where each data point belongs to the cluster with the nearest mean value.

To apply K-means clustering to image clustering, you can treat each image as a high-dimensional vector by representing it with features such as color histograms, pixel intensities, or deep learning embeddings.

It's worth noting that K-means clustering can be sensitive to the initialization of cluster centroids. Different initializations may lead to different clustering results. To overcome this, you can run the algorithm multiple times with different initializations and choose the clustering result with the lowest error or highest evaluation metric value.

Rules

• Programs must be based on C/C++ or Python w/wo window interface

Requirements

Please complete the following requirements

1. The dimension of the feature vector must be more than 10.
2. Use K-means clustering to classify the dataset (Total 10 categories)
3. Calculate the precision rate and recall rate for each category.

Precision & Recall

Calculate the precision rate and recall rate for each category.

• Precision $= \space {{True Positive \over Actual Results} \space or \space{True Positive \over True Positive + False Positive}}$
• Recall $= \space{{True Positive \over Actual Results} \space or \space{True Positive \over True Positive + False Negative}}$

Precision can be seen as a measure of quality, and recall as a measure of quantity. Higher precision means that an algorithm returns more relevant results than irrelevant ones, and high recall means that an algorithm returns most of the relevant results (whether or not irrelevant ones are also returned). For more...

Installation

Download from github

git clone https://github.com/nick8592/Kmeans-Clustering.git

Install reqired dependencies

pip install -r requirements.txt

Example

Add/Remove features in your feature_list in main.py at line 93.
You can place in any kinds of feature in feature_list.

(e.g.) feature_list = [brightness, euler_number, irregularity_ratio, lines, circles]

Run K-means Clustering

python main.py

Real feature method used in main.py please check out utils.py.

You should see the results looks like below format.

Val Labels:
[7 6 0 5 5 7 1 9 8 3 8 5 6 7 9 9 0 2 2 6 4 2 3 4 0 3 7 3 2 0 1 5 9 9 9 7 6
 0 7 9 3 1 6 1 3 2 6 1 1 8 5 5 7 8 5 6 8 5 8 0 8 4 1 3 8 4 3 9 0 4 5 5 6 4
 2 1 1 6 4 4 7 3 0 2 3 8 4 4 0 6 1 9 2 0 2 2 8 7 7 9]
Predicted Labels:
[7 7 3 8 5 7 1 3 3 1 7 1 7 7 0 7 1 2 5 1 7 5 1 1 1 1 7 5 0 5 1 5 1 5 5 3 1
 1 7 1 1 7 8 1 1 1 9 7 1 1 1 8 7 7 5 5 1 5 5 4 5 7 1 7 5 7 1 5 7 1 5 1 7 7
 1 5 7 9 7 7 7 3 4 2 7 1 1 7 3 7 1 8 0 3 8 1 7 7 7 0]
==========================================================================
Features Num: 5
Best random seed: 210
Best random_state: 32
==========================================================================
Class         |  Precision  |  Recall
------------------------------------------
10_flower     |     0.0000  |  0.0000
1_horse       |     0.1935  |  0.6000
2_plane       |     1.0000  |  0.2000
3_rooster     |     0.1429  |  0.1000
4_tree        |     0.0000  |  0.0000
5_sailboat    |     0.2941  |  0.5000
6_motorcycle  |     1.0000  |  0.0000
7_car         |     0.3000  |  0.9000
8_butterfly   |     0.0000  |  0.0000
9_dragonfly   |     0.0000  |  0.0000
------------------------------------------
Total         |     0.2300  |  0.2300

Real feature method used in main.py please check out utils.py

References

Dominate colors

Extract dominant colors of an image using Python
ImageDominantColor
Finding the Most Common Colors in Python
Dominant Color Extraction Dominance and Recoloring

PCA

PCA Using Python: Image Compression