Gumpifier - Our Summer Intern AI Project – Electronic design services with Argon Design Ltd.

Gumpifier - Our Summer Intern AI Project

This summer, we had two excellent undergraduates from the Computer Laboratory at Cambridge University who joined us for an internship - Mohammed Daudali (pictured above) and Patrick Taylor.

We set them the task of creating a web-based AI tool that we called “The Gumpifier”: The Gumpifier will allow the user to input an image of themselves, along with a photo of a background scene, and will automatically insert their picture into the scene in a realistic manner.

We wanted them to explore the use of Deep Learning in image processing and also to think about the idea of User Centred AI. For this we teamed up with Dovetailed, another innovative Cambridge company who specialise in User Experience (UX) design. Hai-Dao, an intern at Dovetailed, led the UX part of the project.

We helped the interns explore ideas over the summer and create a working Gumpifier. You can play with the result on your own images at http://www.gumpifier.com.

Here’s what the site looks like when you have uploaded a pair of images – in this case, a background shot from just outside our office and a picture of Patrick:

As well as placing the person in the image, Gumpifier adds a playful quote from Forrest Gump to the picture. If you don’t want this, just select all of the text and delete it. Here is an example output image for the photos that were uploaded in the previous screenshot:

Congratulations to Hai-Dao, Mohammed and Patrick for a well-spent summer, learning about AI and for creating a fun site for people to use.

You can read their blog post at http://www.gumpifier.com/blogPost.html. We are also open-sourcing the code which is available on our GitHub repository at https://github.com/ArgonDesign/gumpifier.

Argon Design does commercial projects in Deep Learning and combines strong academic mathematical skills for algorithm design with practical experience of training networks and efficient implementations in software and silicon.