DeepDream is an artistic image modification technique that leverages the representations learned by deep CNN code Inception, named after the movie of the same name. We can take any input image and process it to generate trippy pictures, full of algorithmic pareidolia artifacts, bird feathers, dog-like faces, dog eyes—a by-product of the fact that the DeepDream convent was trained on ImageNet, where dog breeds and bird species are vastly over-represented.
The DeepDream algorithm is almost identical to the ConvNet filter visualization technique using gradient ascent, except in a few differences as follows:
- In DeepDream, activation of entire layers is maximized, whereas in visualization only a specific filter is maximized, thus mixing ...
