Engenharia de Computação

URI Permanente para esta coleção

Agora exibindo 1 - 1 de 1

Acesso aberto (Open Access)
Métodos de resolução de problemas de quebra-cabeça visual baseados em aprendizado profundo
(Universidade Federal de São Paulo, 2023-12-12) Taciano, Miguel Silva [UNIFESP]; Faria, Fabio [UNIFESP]; http://lattes.cnpq.br/3828728429230356; http://lattes.cnpq.br/3494257311247402
The traditional Jigsaw Puzzle is a challenging task performed by humans, mainly due to its hardness and being proven to be an NP-Complete problem. Even so, recent efforts show better performance in this task using different methods involving complex computer vision and machine learning techniques. In this sense, this thesis proposes new methods based on the use of semantic segmentation (SS) and deep learning, alone and combined, with the tasked objective to solve jigsaw puzzles (visual puzzles) in different training scenarios, both with a known middle and an unknown middle configuration. To the best of our knowledge, this is the first work in the literature that uses SS for the target application. In the performed experiments, it was possible to demonstrate that our proposed methods successfully obtained excellent results when compared with other methods existing in the literature for 3 × 3 puzzle-solving tasks.