Deep Dreams implementation using Chainer

Requirements

• Python 2.7
• Chainer 2.0.0
• Cupy 1.0.0
• Pillow 3.1.0

Usage

Download googlenet caffe model

Download bvlc_googlenet.caffemodel from https://github.com/BVLC/caffe/tree/master/models/bvlc_googlenet, and put it into repository root.

Run script

$ python deepdream.py -g 0 -i sample.png -o out_image

• -g n: GPU device index, or -1 to run without GPU.
• -i file_path: Input image file path.
• -o dir_path: Output image directory path.
• -m file_path: Caffe model file path. Default is "bvlc_googlenet.caffemodel".
• -s size: Output image size. Image will be zoomed so that the larger side length equals this value.
• --iter n: Number of iterations. Default is 100.

Generated image sample

Original image
<img src="https://raw.githubusercontent.com/dsanno/chainer-deepdream/master/image/original.png" width="484px alt="generated image sample">
Generated after 20 iterations
<img src="https://raw.githubusercontent.com/dsanno/chainer-deepdream/master/image/generated.png" width="484px alt="generated image sample">