Multi-Model-Classification

Project #1

A few days ago, I saw a meme on Instagram. The meme was a bunch of similar looking images arranged in a matrix. I thought it would be a nice little project, if I could train a simple CNN model to predict each images. Above that, I train a Main model which has the ability to predict with class each images belongs to.

Working:

  1. Image is fed
  2. Main model predicts which class it belong to, out of 8 classes:
    a. Chihuahua V Muffin
    b. Kitten V Ice Cream
    c. Labradoodle V Fried Chicken
    d. Parrot V Guacamole
    e. Puppy V Bagel
    f. Sheepdog V Mop
    g. Shiva Inu V Marshmallow
    h. Sloth V Croissant
  3. Then, individually trained models, classifies it in binary.
  4. Class and label is printed, along with the image.

Original Images:

Full Disclosure, the modela are pretty bad....given the very few images available to train.
But overall, it works amazingly.


Example: