Machine learning for images

Sobre este projeto

Aberto

1 - OBJECTIVE: Generate new images of typographic characters from a predefined database (attached).
2 – DEVELOPMENT: Three parallel approaches to solve the same problem.
important: choose only one of the following options to respond to the offer
solution 1: development with gans (generative adversarial networks)
solution 2: development with transformers
solution 3: list of commercial tools to leverage the database
3 – relevant aspects
comparative study of training and quality between tool 1 and tool 2.
Architecture focused on facilitating the development of future versions for other images.
Ability to run training on a computer without an internet connection.
Relationship between hardware power and training time and image creation.
Size of the database and minimum image quality.
Need for manual labeling of images in the database.

Categoria TI e Programação
Subcategoria Data Science
Tamanho do projeto Pequeño
Isso é um projeto ou uma posição de trabalho? Um projeto
Disponibilidade requerida Conforme necessário

Prazo de Entrega: Não estabelecido

Habilidades necessárias

Database Python Machine Learning Pytorch Tensorflow Jupyter Notebook

Sobre este projeto

it-programming / data-science-1

Aberto

Outro projetos publicados por R. U.