Last year, I worked on a photoshop like tool for ecommerce, where we can fit garments on models - https://www.youtube.com/watch?v=WUJizLU6ALo . Forgive me for the poor video, it was quickly prepared with some old data and I did not spend time improving the output.<p>I want to pitch this product to adobe / on1 / etc; it can become a cool image processing tool, specially meant for ecommerce catalog generation, etc. Can you suggest me the companies that would be interested in hiring me to build this?<p>PS... internal details of the implementation and techniques are left out in the video (because they are learnings over months of experimentation).
First off, great work. This is impressive stuff. You should definitely put up a page describing some of the technical details, this is hacker news after all and a lot of us would be interested.<p>I think you have something and here's why.<p>Awhile back I started to devise a new kind of advertisement system based on this same concept. My starting point was the research done by Kota Yamaguchi. <a href="http://vision.is.tohoku.ac.jp/~kyamagu/research/clothing_parsing/" rel="nofollow">http://vision.is.tohoku.ac.jp/~kyamagu/research/clothing_par...</a><p>Although most of that work was done in matlab, I was able to translate and extend the core ideas into something practical. In this type of domain, clothing segmentation and pose estimation is key.<p>I don't know how your images and data are set up, but with my software you could feed it a random facebook image, feed it some images of different types of clothing and the output would be that person wearing an entirely different outfit. Hopefully you can see how this can easily become a big business.<p>Anyways, I never released anything publicly because the distortion and displacement algorithms were never perfected (Albeit with hilarious results... Jeans became capris, shorts became miniskirts that kind of thing). Given the time I'd love to jump back into this.<p>In any case, I'm interested in the technical details and the approach you took.