When classifying data by KNN, the classes are calculated by using the distance between datapoints. For example, the case of real estate, where the x-axis is the price and the y-axis the size of real estate. But what are the values of axis when instead of numbers there are pictures? By which category are the distances calculated in this case?
Lets say its a gray scale image. And of size 224*224 than you have 50176 possible pixel values. So you can plot on x axis widt=224 and on y heigth=224.
Lets say they are coloured. Than you have more chanels=dimensions. In that case you perform clustering in higher dimensional space