2026-05-31T07:33:16Zhttps://riuma.uma.es/rest/oai/request

oai:riuma.uma.es:10630/385052026-02-03T11:17:40Zcom_10630_2254col_10630_37953

Enhanced generation of automatically labelled image segmentation datasets by advanced style interpreter deep architectures. Pacheco dos Santos Lima Junior, Marcos Sergio López-Rubio, Ezequiel Ortiz-de-Lazcano-Lobato, Juan Miguel Fernández-Rodríguez, Jose David Redes neuronales artificiales Aprendizaje automático (Inteligencia artificial) Large image datasets with annotated pixel-level semantics are necessary to train and evaluate supervised deep-learning models. These datasets are very expensive in terms of the human effort required to build them. Still, recent developments such as DatasetGAN open the possibility of leveraging generative systems to automatically synthesise massive amounts of images along with pixel-level information. This work analyses DatasetGAN and proposes a novel architecture that utilises the semantic information of neighbouring pixels to achieve significantly better performance. Additionally, the overfitting observed in the original architecture is thoroughly investigated, and modifications are proposed to mitigate it. Furthermore, the implementation has been redesigned to greatly reduce the memory requirements of DatasetGAN, and a comprehensive study of the impact of the number of classes in the segmentation task is presented. 2025-05-07T07:20:09Z 2025-05-07T07:20:09Z 2025 journal article Marcos Sergio Pacheco dos Santos Lima, Ezequiel López-Rubio, Juan Miguel Ortiz-de-Lazcano-Lobato, José David Fernández-Rodríguez, Enhanced generation of automatically labelled image segmentation datasets by advanced style interpreter deep architectures, Pattern Recognition Letters, Volume 193, 2025, Pages 101-107 https://hdl.handle.net/10630/38505 10.1016/j.patrec.2025.04.021 eng http://creativecommons.org/licenses/by-nc/4.0/ open access Attribution-NonCommercial 4.0 Internacional Elsevier