June 2019
Intermediate to advanced
348 pages
8h 28m
English
In this subsection, we will train a cascade classifier for a less rigid object: dog faces.
We will use the dataset from http://vision.stanford.edu/aditya86/ImageNetDogs/. This dataset contains 20,580 images of dogs, divided into 120 categories, and each category is a dog breed. Let's download and unpack the tarball of the images:
$ curl -O http://vision.stanford.edu/aditya86/ImageNetDogs/images.tar$ tar xvf images.tar# output omitted$ ls Images/n02085620-Chihuahua n02091635-otterhoundn02097298-Scotch_terrier n02104365-schipperken02109525-Saint_Bernard n02085782-Japanese_spaniel# output truncated
We will set the faces of the Boston Bull breed as our target. There are 182 images of Boston Bulls in the Images/n02096585-Boston_bull ...
Read now
Unlock full access