top of page


Computer Vision, Machine Learning, 2017

RUIN is an image processing research experiment with google deepdream algorithm. The learning algorithm of this project was trained with thousands of photography works I took during my stay in Japan at 2012 and was used to visualize how machine sees the world via the lens of my cameras. The model was later applied onto some of my Fukushima photo series, which creates a dialogue between the machine and the technological disaster.

Deep Dream Fukushima

When machines were equipped with the ability and "knowledge" to see this world, how will it look upon humanity and its technological landscape?


Technical Details

Training framework: Caffe, VS 2015, CUDA 8.0, cuDNN v5, Python + Jupyter Notebook

Training Duration: 60 - 150 hours


Photo Documentation

Other Projects

All Images Copyright © Mengyu Chen | Media Art, Design, Technology

bottom of page