Technology
Thu Jul 23,2015 10:00 am
A Half-Million Images Have Been Uploaded and Processed Through Deep Dream Web Interface on Over 600 CPU Cores
Psychic VR Lab
Psychic VR Lab launches Deep Dream Web Interface, which enables anyone to easily experience Deepdream. Within two weeks, a half-million photos have been uploaded. Currently more than 600 CPU cores are running and processing 50,000 images per day.
July 24, 2015
(Tokyo)- Psychic VR Lab launches Deep Dream Web Interface (http://psychic-vr-lab.com/deepdream/), which enables users to upload their own photos to produce dreamy images. Now the site becomes one of the most popular DeepDream image generators. Deepdream is a code developed by Google. This algorithm allows us to process ordinary images into “nightmares” using Artificial Intelligence program that looks for similarity in patterns.
Originally, this service was started with a few computers that were not used in the lab to examine how DeepDream would work. To our surprise, immediately after the service launches, thousands of images were uploaded from all over the world. Deepdream requires computationally intensive algorithm and it takes 10 to 20 minutes to process one photo, so inevitably there was a long waiting line of images to be processed.
Number of images uploaded to the web surged as the images generated became popular on social networks, so we recruited sponsors. Host Duplex, LLC has generously provided us with 244 CPU cores, which greatly improved processing capability.
Simultaneously, Psychic VR Lab has developed a system that uses a part of the individual PC resources of site visitors for Deep Learning algorithm. Now many of the users install the system on their PC, and images that are uploaded by those who cooperated us are processed on a priority basis. Thanks to their help, processing capability has been improved further.
Currently Deep Dream Web Interface is running on over 600 CPU cores all over the world, and it is capable of processing 50,000 images per day. Psychic VR Lab hopes to utilize this Deep Learning Grid Network for other research purposes than image processing.
About Psychic VR Lab
Psychic VR Lab is established as a R&D division of Ozmiq, Inc. and Intronworks Ltd. Our research focuses on human augmentation and extraordinary experiences with Virtual Reality. We are currently working with Hokkaido University conducting a research project on sense manipulation using visual information with the aim of medical application. We believe that our research on Virtual Reality and Mixed Reality will change how people live their lives and improve quality of life.
Contact
Masahiro Yamaguchi, CEO
info@psychic-vr-lab.com
(Tokyo)- Psychic VR Lab launches Deep Dream Web Interface (http://psychic-vr-lab.com/deepdream/), which enables users to upload their own photos to produce dreamy images. Now the site becomes one of the most popular DeepDream image generators. Deepdream is a code developed by Google. This algorithm allows us to process ordinary images into “nightmares” using Artificial Intelligence program that looks for similarity in patterns.
Originally, this service was started with a few computers that were not used in the lab to examine how DeepDream would work. To our surprise, immediately after the service launches, thousands of images were uploaded from all over the world. Deepdream requires computationally intensive algorithm and it takes 10 to 20 minutes to process one photo, so inevitably there was a long waiting line of images to be processed.
Number of images uploaded to the web surged as the images generated became popular on social networks, so we recruited sponsors. Host Duplex, LLC has generously provided us with 244 CPU cores, which greatly improved processing capability.
Simultaneously, Psychic VR Lab has developed a system that uses a part of the individual PC resources of site visitors for Deep Learning algorithm. Now many of the users install the system on their PC, and images that are uploaded by those who cooperated us are processed on a priority basis. Thanks to their help, processing capability has been improved further.
Currently Deep Dream Web Interface is running on over 600 CPU cores all over the world, and it is capable of processing 50,000 images per day. Psychic VR Lab hopes to utilize this Deep Learning Grid Network for other research purposes than image processing.
About Psychic VR Lab
Psychic VR Lab is established as a R&D division of Ozmiq, Inc. and Intronworks Ltd. Our research focuses on human augmentation and extraordinary experiences with Virtual Reality. We are currently working with Hokkaido University conducting a research project on sense manipulation using visual information with the aim of medical application. We believe that our research on Virtual Reality and Mixed Reality will change how people live their lives and improve quality of life.
Contact
Masahiro Yamaguchi, CEO
info@psychic-vr-lab.com
.jpeg)
