Guided Deep Dreaming | with control images | gradient ascent | at layers of VGG16 | pytorch
Guided Deep Dreaming with a few different guide images with gradient ascent at different layers, using VGG16 pretrained weights with pytorch / python
As can be seen, performing gradient ascent with guided features (starting with pretrained VGG16 weights) amplifies simpler edge-like features at initial layers and more complex patterns deeper layers, respectively
(#pytorch #computervision #deeplearning #python #pythonprogramming)