4

There are several generative models that have been proposed before or roughly at the same time of the GAN (2014). For example, the deep Boltzman machine (2009), deep generative stochastic network (2014) or variational auto-encoder (2014).


3

The paper StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks should provide the answers to your questions. Here's an excerpt from the abstract of the paper. Synthesizing photo-realistic images from text descriptions is a challenging problem in computer vision and has many practical applications. Samples generated ...


3

The purpose of Reinforcement Learning is to maximize some notion of cumulative reward, leading me to the point (1) : as far as I understand, there is no timesteps in your problem and the "reward" is immediate. Thus, I don't think reinforcement is suitable here. On an other hand, in supervised learning, linear regression is the task of approximating a ...


2

I'm not an expert on that so you could probably get a better answer. I'm not sure to understand what you're looking for. Are the couple of images about the same thing? Like pictures of cats and you want to generate a new cat based on these pictures? If that's what you want, you could probably take a look at Generative Adversarial Network (GAN) : ...


2

[Answering my own question after 5 months of studying VAE models] The point of the MMD-VAE or InfoVAE is not exactly to emphasise on the visual quality of generated samples. It is to preserve greater amount of information through the encoding process. The MMD formulation stems from introducing a mutual coefficient factor into the Evidence Lower BOund (ELBO) ...


2

The only disadvantage and difference between these generative models and the method you describe, is the input. You describe inputting tags, where as for a GAN, or VAE, the generation segment of the model takes in some representation of a probability distribution. For a GAN, it's mostly random noise, and for a VAE it is some latent space (see nbros answer). ...


2

This would likely suffer from the blurry image problem that autoencoders are known to suffer from. See also here. On the other hand, using GAN's to sharpen your images doesn't seem particularly helpful since you seem to be lookng for a way to rotate general images, not ones of a specific domain. Moreover, there's almost certainly going to be some loss It ...


1

In computer vision, the problem of filling missing parts of an image is called image inpainting; the subtask of filling the surroundings is called image outpainting in [1], which is your problem. The methods for solving the image outpainting problem are not mature according to the pre-print paper Image Outpainting and Harmonization using Generative ...


1

Evaluating synthetically generated images is challenging and an active area of research. The problem is that the "how natural is an image"-task is not well-defined and subjective. To evaluate generated images we can define two abstract properties: fidelity and diversity, as we want to generate not only a single high-quality image, but also ...


1

DALL·E is a 12-billion parameter version of GPT-3 trained to generate images from text descriptions should be the same data they used to train the GPT-3


1

Assuming that the image is blank everywhere but where the face is drawn... The first step is to scale the image to the mask. That doesn't require a detailed explanation here as it is too trivial a problem. Second, rotate the image by 90 degrees three times and save each one. Third, for the four versions of the image (the original and three rotations), do ...


1

Just use IMGAug library for Applying the 'zoom' augmentation on the images and a convnet (or even MLP) would have no problem in this task. Zooming on the image would be more than enough to use as, as long as your zoom is of sufficient power. So apply "zoom" of the same factor on all images, make them bigger, and Convolutional Neural Networks would ...


1

Instead of NNs, you can use RANSAC algorithm to calculate homography matrix, but first you need to find feature points. However, if your images are blob-like, you may not get such a successful results. Here some presentations for better understanding: cs.umd notes and csail.mit notes (Also, there might be better image processing tools.)


1

GANs generally produce better photo-realistic images but can be difficult to work with. Conversely, VAEs are easier to train but don’t usually give the best results. I recommend picking VAEs if you don’t have a lot of time to experiment with GANs and photorealism isn’t paramount. There are exceptions such as Google’s VQ-VAE 2 which can compete with GANs for ...


1

Some excerpts from Nutella 'Hired' an Algorithm to Design New Jars. And It Was a Sell-Out Success: The "algorithm" is called HP Mosaic and is included free in HP SmartStream Designer for HP printers. More about how the algorithm works here: https://www.linkedin.com/pulse/hp-mosaic-20-steven-chow HP Mosaic takes the vector PDF file as input (also known ...


1

I'd challenge your assertion somewhat that the generated images of other categories are of much worse quality than the faces! Take the bikes on transparent / solid backgrounds they look great! Where the images fail a bit is with the more complex pictures which have a lot of elements where element bleed (covers bleeding into the floor, etc.) occurs. This is ...


1

Generative Adversarial Networks, basically boil down to a combination of a generic Generator and a Discriminator trying to beat each other, so that the generator tries to generate much better images (usually from noise) and discriminator becomes much better at classification. So, no it is not just suited for only synthesis high quality human face synthesis ...


1

I will only focus on the VAE because I am more familiar with it, but the explanations may also apply to the GAN and other generative models. In the case of the VAE, you train a neural network not only to generate images but to represent them compactly in a so-called latent space, so you train the VAE to do dimensionality reduction. More precisely, the VAE ...


1

I believe you may want to use a Sum Product Network for this task. SPNs are the state-of-the-art approach for face completion, and there are several more recent papers on this topic since the original above. Importantly, the SPN paper also covers other approaches that work well for this task. If lower-resolution results are acceptable for your task, PCA ...


1

The model (that I know of) which most resembles your description is the auto-encoder, which is trained to learn a compact representation (a vector) of the input, which can later be used to reconstruct the original input. In a certain way, this compact representation (implicitly) encodes the most important features of the input. In particular, you may be ...


1

I wouldn't really consider what they are doing as AI - they are using a script that intelligently overlaps various images of existing people in order to create a new face. Animating those images isn't impossible - essentially extrapolation + additional "real" images will be used to know what the face would like from all angles and in all states (happy, sad, ...


Only top voted, non community-wiki answers of a minimum length are eligible