AttnGAN Neural Network to Draw Strange Pictures

The neural network is good at drawing birds by the text description, but bad everything else
20 August 2018   1509

The author of the AI Weirdness blog Janelle Shane had discovered the generative-controversial neural network called AttnGAN, which is trained to draw images on the text description. The problem is that it requires too accurately defined picture parameters and sometimes can not determine the boundaries of objects.

Janelle notes that, while the neural network was trained on a narrow set of data in the form of birds, it obtained nice images:

AttnGAN
AttnGAN

However, when the creators trained it on a dataset that included pictures from sheep to shopping centers, it could not create a meaningful image in a similar way. The author of AI Weirdness believes that the error lies in too wide a set of initial data, in which AttnGAN could not select the appropriate instances:

AttnGAN
AttnGAN

In addition, it somehow has a problem with determining the correct number of holes on the human face. Developers AttnGAN added to the control dataset person celebrities to create photorealistic portraits, but the neural network couldn't do that:

AttnGAN
AttnGAN

Additionally, neural network is real bad at displaying animals:

AttnGAN
AttnGAN

Janelle Shane calls the project AttnGAN "Visual Chatbot on the contrary." This chat bot analyzes the image that the user sends and describes it, often implausibly.

Nvidia to Open SPADE Source Code

SPADE machine learning system creates realistic landscapes based on rough human sketches
15 April 2019   875

NVIDIA has released the source code for the SPADE machine learning system (GauGAN), which allows for the synthesis of realistic landscapes based on rough sketches, as well as training models associated with the project. The system was demonstrated in March at the GTC 2019 conference, but the code was published only yesterday. The developments are open under the non-free license CC BY-NC-SA 4.0 (Creative Commons Attribution-NonCommercial-ShareAlike 4.0), allowing use only for non-commercial purposes. The code is written in Python using the PyTorch framework.

Sketches are drawn up in the form of a segmented map that determines the placement of exemplary objects on the scene. The nature of the generated objects is set using color labels. For example, a blue fill turns into sky, blue into water, dark green into trees, light green into grass, light brown into stones, dark brown into mountains, gray into snow, a brown line into a road, and a blue line into the river. Additionally, based on the choice of reference images, the overall style of the composition and the time of day are determined. The proposed tool for creating virtual worlds can be useful to a wide range of specialists, from architects and urban planners to game developers and landscape designers.

Objects are synthesized by a generative-adversarial neural network (GAN), which, based on a schematic segmented map, creates realistic images by borrowing parts from a model previously trained on several million photographs. In contrast to the previously developed systems of image synthesis, the proposed method is based on the use of adaptive spatial transformation followed by transformation based on machine learning. Processing a segmented map instead of semantic markup allows you to achieve an exact match of the result and control the style.

To achieve realism, two competing neural networks are used: the generator and the discriminator (Discriminator). The generator generates images based on mixing elements of real photos, and the discriminator identifies possible deviations from real images. As a result, a feedback is formed, on the basis of which the generator begins to assemble more and more qualitative samples, until the discriminator ceases to distinguish them from the real ones.