Fake Dogs

Our AI was trained to create dogs, using a database of 20,000 of photos of real dogs. To be clear, these dogs do not exist. They are not real.

827


The dog images on this page are created using a Generative adversarial network (GAN), which is a machine learning configuration where in this case one neural network generates fake dogs and another neural network tries to detect which are real and which are fake. Through this process, over time the first neural network learns to get better at producing the fake dogs, in order to fool the second neural network.

We created 50 fun artificial intelligence (AI) pages you can use for free:


  • Coming Soon - BoredGPT - Like ChatGPT, but more acurate and more powerful.
  • Virtual Pets - Adopt a free virtual dog, cat, parrot, platypus, kangaroo, pig, turtle, koala bear, monkey, or horse.
  • Coming Soon - Rock Paper Scissors - Use the camera on your phone or PC to play against the AI.
  • Coming Soon - AI Stock Market Predictions - We use machine learning to predict the prices of stocks and crypto coins.
  • Coming Soon - Tarot Card Readings - Find out your future using AI.
  • Coming Soon - Text Adventure Game - Interactive fiction written by AI.
  • Coming Soon - Guess the Jellybeans - Can you guess better than our AI how many jelly beans are in the jar?


Back To BoredHumans.com Main Page



Privacy Policy
(We do not save your info, sell your personal data, or do anything else to violate your privacy)

Copyright: The design of this site is Copyright 2023 by BoredHumans.com.
BoredHumans.com is owned by:
Impulse Communications, Inc.
9450 SW Gemini Dr. #56742
Beaverton, OR 97008
Email: eric@boredhumans.com

Our dog image generator was made using the 1st Place Solution from the Kaggle Generative Dog Images contest, which is licensed Apache 2.0. It uses a GAN (Generative Adversarial Network) create new images that it thinks are real looking. The database of dogs that was used to train the machine learning model was Stanford Dogs Dataset. The output of the ML model is then filtered using a modified version of Dog Face Detector to delete the dog images that do not contain faces (around 2/3 of them) because those are usually no good.