Step 1

Isolate the butterflies from the background using the given segmentation masks.

Step 2 & 3

Calculate the average intensity of red, green and blue color per image.
Calculate the average color intensity per class.

Step 4

Compare the euclidean distances between the input image and all of the 10 classes.

Step 5 - Results

The smallest distance represents the predicted class of the input image.

Classification accuracy

Here we are getting an accuracy of about 42%. The algorithm works very well on some species, because of their specific color.