Can PCA be used on dimensionality reduction of two images, where one is the inverted image of the other?

C1W4_Assignment

Hi, I am new to machine learning, and I would like to know whether the PCA works for the 2 same images, but one is an inverted image of the other.

In the assignment all of the cats’ images appear to be upright, what if we have an image of the same cat but is rotated 180 degrees, and fed it to the algorithm?

In the assignment, it has been mentioned that while performing analysis after reducing rows, we can see two similar images plotted very close on the graph? Does the algorithm assume the inverted images as different even though they are same?

This seems like a good opportunity for you to perform some experiments.

Please report back your results.

1 Like