DeepDream
Deep dream implementation with inception model
DeepDream is an artistic image-modification technique that uses the representations learned by convolutional neural networks. It was first released by Google in the summer of 2015, as an implementation written using the Caffe deep-learning library (this was several months before the first public release of TensorFlow).4 It quickly became an internet sensation thanks to the trippy pictures it could generate, full of algorithmic pareidolia artifacts, bird feathers, and dog eyes—a byproduct of the fact that the DeepDream convnet was trained on ImageNet, where dog breeds and bird species are vastly overrepresented.
What is really happening
1- With DeepDream, you try to maximize the activation of entire layers rather than that of a specific filter, thus mixing together visualizations of large numbers of features at once.
2- You start not from blank, slightly noisy input, but rather from an existing image—thus the resulting effects latch on to preexisting visual patterns, distorting elements of the image in a somewhat artistic fashion.
3- The input images are processed at different scales (called octaves), which improves the quality of the visualizations.
Code describtion
After importing the packages we will be using
import numpy as np
import PIL.Image
import cv2
import os
from deepdreamer import model, load_image, recursive_optimize #import some functions from the other file
next line determine that we are going to use the contribution of only layer 3 loss from the inception model(responsible for detection of edges) x_size and y_size are just the dimensions of the picture i picked inside the loop load the image and zoom to a certain ratio
x_trim = 2
y_trim = 1
img_result = img_result[0+x_trim:y_size-y_trim,0+y_trim:x_size-x_trim]
then resize it to its original size tune the colors in the image img_result[:,:,0] +=2 img_result[:,:,1] +=2 img_result[:,:,2] +=2 you can tune it we any value you like and then restrict the values of the tensor from 0 to 255, then call the recersive function (described inside the other file) perfome same steps as before and then save it, the loop will run 50 times generating 50 pictures and then you can easily put them as frames together resulting in a video and the code for that provided ^