![]() Photo Repair is one of several technologies that have been licensed from DeOldify and integrated into the MyHeritage platform. The technology for repairing scratched and damaged photos was licensed by MyHeritage exclusively from DeOldify, created by deep learning experts Jason Antic and Dana Kelley. Photo Repair is effortless and immediate: it is suggested only when it is needed, and is activated by the user with a single click. MyHeritage is the only company to offer a complete suite of features for colorizing, enhancing, animating, and now repairing historical photos, all of which produce exceptional results. Photo Repair takes photos that have deteriorated over the years and corrects the damage, making flaws disappear as if by magic. MyHeritage, the leading global service for discovering your past and empowering your future, announced today the release of Photo Repair, a powerful new feature that automatically fixes scratches, tears, holes, stains, and other damage on historical photos. The best way to support me is by following me on Medium.New AI-based feature reinforces MyHeritage’s position as the market leader for storing and improving historical photos.If you like my work and want to support me, I’d greatly appreciate it if you follow me on my social media channels: It is on MyHeritage’s website and is paid to use. Finally, the third link is the most advanced version of DeOldify if you are looking for the best results. Then, you can find a free API on DeepAI using DeOldify where you can simply click and try yourself. At first, there’s the GitHub link with a complete detailed explanation of the technique and even google colab tutorials to use it yourself. You can find three things in the references below. Right now, there is no complete explanation of how this works, but the author is currently working on a paper about DeOldify, where he will further investigate why and how his technique, previously found only by trials and errors work. Where the generator network in the GAN training is the U-Net architecture. The whole architecture uses a basic ResNet backbone on a U-Net. This is a type of data augmentation that can be performed on the training images to improve the results and resistance to noisy inputs, using the same technique as style transfer, where the noise would be the style of the image we want to copy and can be applied more or less to the transformation. Gaussian noise is also randomly applied to images to generate fake noise during training. Then, it only needs a short amount of this typical generator-discriminator GAN training to optimize the “realism” of the generated pictures. That way, the model is already pretty good at colorizing an image before training the complete GAN architecture. This is done by training the generator like a regular deep network’s architecture, such as ResNet. Instead, he pre-trains the generator to make it already more powerful, fast, and reliable using a regular loss function. His new method, which he calls the “NoGan”, provides the same benefits of this usual GAN training while having to spend way less time training the GAN architecture, which is typically pretty heavy in computation time. If this was just completely abstract to you, I invite you to watch the video I made about GANs: ![]() Typically, GAN training works by both training the discriminator and generator at the same time, where the generator starts by being completely random and improves over time to fool the discriminator, which tries to tell if the image is generated or real. It uses a new type of GAN training method called NoGAN that he developed himself to solve the main problems that appeared when training using a normal adversarial network architecture composed of a discriminator and a generator. ![]() It is now the state of the art way to colorize black and white images, and everything is open-sourced, but we will get back to this in a bit.įirst, let’s see how he achieved that. It was developed and is still getting updated by only one person Jason Antic. 3 min read Watch the video and support me on YouTube!ĭeOldify is a technique to colorize and restore old black and white images or even film footage.
0 Comments
Leave a Reply. |