Centre de documentation
scientifique

Prototype-guided multitask adversarial network for cross-domain LiDAR point clouds semantic segmentation / Zhimin Yuan in IEEE Transactions on geoscience and remote sensing, vol 61 n° 1 (January 2023)  

Public [article]  inIEEE Transactions on geoscience and remote sensing > vol 61 n° 1 (January 2023) . - n° 5700613 Titre : Prototype-guided multitask adversarial network for cross-domain LiDAR point clouds semantic segmentation Type de document : Article/Communication Auteurs : Zhimin Yuan, Auteur ; Ming Cheng, Auteur ; Wankang Zeng, Auteur ; et al., Auteur Année de publication : 2023 Article en page(s) : n° 5700613 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes IGN] alignement des données [Termes IGN] apprentissage non-dirigé [Termes IGN] compression de données [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] extraction de traits caractéristiques [Termes IGN] réseau antagoniste génératif [Termes IGN] segmentation sémantique [Termes IGN] semis de points Résumé : (auteur) Unsupervised domain adaptation (UDA) segmentation aims to leverage labeled source data to make accurate predictions on unlabeled target data. The key is to make the segmentation network learn domain-invariant representations. In this work, we propose a prototype-guided multitask adversarial network (PMAN) to achieve this. First, we propose an intensity-aware segmentation network (IAS-Net) that leverages the private intensity information of target data to substantially facilitate feature learning of the target domain. Second, the category-level cross-domain feature alignment strategy is introduced to flee the side effects of global feature alignment. It employs the prototype (class centroid) and includes two essential operations: 1) build an auxiliary nonparametric classifier to evaluate the semantic alignment degree of each point based on the prediction consistency between the main and auxiliary classifiers and 2) introduce two class-conditional point-to-prototype learning objectives for better alignment. One is to explicitly perform category-level feature alignment in a progressive manner, and the other aims to shape the source feature representation to be discriminative. Extensive experiments reveal that our PMAN outperforms state-of-the-art results on two benchmark datasets. Numéro de notice : A2023-118 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1109/TGRS.2023.3234542 Date de publication en ligne : 05/01/2023 En ligne : https://doi.org/10.1109/TGRS.2023.3234542 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=102489 [article]