July 21, 2020 Vladyslav Kyrychenko

Photo Augmentation tool that with one touch will make an ideal picture out of a good one

Problem

There is no limit to perfection, but what if your aim is to get as close to it as possible? Our client’s company was created to make enhancing photos both simple and stunningly. But sometimes even perfect tools need an outside perspective. Despite there are plenty of other image editing tools, all of them need certain expertise and fine-tuning.

But our client intended to develop an automatized tool bordering artificial intelligence. That is how our collaboration began. Going forward let’s say that it was as bright and creative as photos enhanced in app.

Task

Our client needed an experienced team who would participate in a computer vision challengeable project. Well, obviously it was us. So the matter was to:

  • create a system to automatically and professional enhance photos;
  • build a model for business denoising;
  • develop a model for image inpainting;
  • create a system for image style transfer.

Summarizing the points, final models and systems were supposed to help users enhance photos quickly and easily. Still, it was not about accustomed photo filters, but about an adjusting system of photos correction bordering artificial intelligence.

Solution

If you are ever planning to do something similar, here is a list of technologies that are ideally suited for this purpose. We have applied it, tested and were satisfied with the result, so check & use it without any hesitation:

  • Python;
  • C++;
  • TensorFlow;
  • Pytorch;
  • Caffe.

Our expertise in deep learning and computer vision has also done a good job for us because the key feature of application is not just photos correction but making these processes smart to get users’ job done as close to perfection as possible.

Result

Since the task was creative, the workflows were also full of creativity. Here is the result:

  • We built several models to enhance photos and deployed from GPU to CPU with little time leaks. We also deployed this solution to desktop to make app more usable and to ease working processes.
  • The solution for image denoising with one of the best result available on the market was created.
  • We also developed a model to inpaint objects up to 256×256.
  • Finally, we created the system for image style transfer and deployed it to iOS/Android applications.

Utilizing photo editors was never as simple, fast and resultative as now due to the bright idea, precise work of experts and benevolent communication between all engaged over the whole period of collaboration. Hope that either client itself and this case study would be beneficial to you whatever you are about to pull off.

LET'S WORK TOGETHER

We are helping Startups and mature Businesses accelerate growth by building development centers.

Ukraine

77 Sechevyh Strelciv,
Kyiv, Ukraine

12 Mechnikova,
Odessa, Ukraine

hello@lemapp.tech

+380 63 418 79 79

Lemapp

© 2017 – 2020
contact-section