Hello, I need someone tat can do this assigment.
CIFAR 10 has 10 categories and
60000 total images, with 6000 images per class. For HW-1, let us just take the
first 100 images from the data set per class and form a data set of 10K images
with 10 labels. Objectives are, a) [10pts] compute a HSV kmeans model with
total number of entry K=64, b) for each image compute a color histogram, c) use
Euclidean, and KL Distances to measure the similarity between two images, i.e,
have a matlab/python function,[d]=getHSVDistance(im1, im2, t); where t is the
table from Kmeans. d) randomly select 100 images from the data set, and find
the 10-nearest neighbors and find out the ratio of matching labels.
0 comments