reply

Posted

print

favorite this post Torch Developer to Deploy Neural Network Art Code to AWS from Github hide this posting unhide

compensation: I'm looking to hire the best of the best. Pay will reflect this accordingly.

Hello,

I am looking for a Torch developer to install this artificial intelligence painting software I found https://github.com/jcjohnson/neural-style onto an Amazon Web Services (AWS) server (plus add a basic GUI if the software does not come with one) so that I can use it to do things such as combine 1.jpg with 2.jpg to generate 3.jpg (see this link https://www.dropbox.com/sh/si4gdi9rfrty0mu/AADHxSf2ypt3MFq4ysEEdC4ua?dl=0 for 1.jpg, 2.jpg and 3.jpg)

I'm guessing I'd access the Amazon server to use the software via some sort of remote desktop solution you help me setup on my Apple laptop, but am open to whatever solution you think is simplest.

It would be great if I could run the software on my laptop itself without having to use a remote Amazon server, but I don't think this is possible based on what I read on the GitHub link above. If it is possible, I will gladly hire you to set it up on my laptop accordingly (and probably Amazon as well for very fast processing). See 4.jpg in this link https://www.dropbox.com/sh/si4gdi9rfrty0mu/AADHxSf2ypt3MFq4ysEEdC4ua?dl=0 for details on my laptop's hardware.

Couple other notes:

1. The original images I am using are around 8k by 8k (64 million pixels). The resulting image the software generates also needs to be around 8k by 8k (64 million pixels). In other words, the resulting image needs to keep the same pixel quality/resolution as the original images. This is because the resulting image in intended to be printed at sizes ranging from the size of a living room poster to the size of a billboard covering a large building in Times Square.

2. I am not 100% tied to this software https://github.com/jcjohnson/neural-style and am open to using anything similar if you feel it can accomplish the same goal and will work better for my needs (and/or gives a higher quality resulting image). My requirements to an alternative to https://github.com/jcjohnson/neural-style are as follows:

A. Can combine 1.jpg and 2.jpg to generate 3.jpg (see this link https://www.dropbox.com/sh/si4gdi9rfrty0mu/AADHxSf2ypt3MFq4ysEEdC4ua?dl=02 for 1.jpg, 2.jpg and 3.jpg).
B. Is pixel agnostic (or can at least process 8k by 8k (64 million) pixel photographs). In other words, it should be able to handle any amount of pixels so long the hardware it is running on can handle them.
C. Ideally, does not cost a fee each time I use it.
D. Would be nice, but completely not required, if it was a Photoshop plugin or had some sort of Photoshop integration.

3. This is a one time freelance project in which you will be working remotely. However, I often have software projects similar to this and am looking to build a long term working relationship with a developer.

Please include the following in your response:
-Your resume and/or Linkedin profile.
-How much it would cost for you to do the project.

Thanks and I look forward to your thoughts!
  • do NOT contact me with unsolicited services or offers

post id: 6792141493

posted:

updated:

best of [?]