Synthetic data find application in computer vision tasks for a long time. Limited performance of algorithms trained solely on synthetic data has been approached with domain adaptation techniques such as the ones based on generative adversarial framework [1]. In this work we demonstrate how using adversarial training alone can introduce semantic inconsistencies in refined images. We suggest leveraging available semantic labels from target domain using naive re-sampling approach alongside with adversarial loss. We also show that aforementioned strategy improves quality of translated images of underlying method and their usability for the semantic segmentation task in the context of autonomous driving. This method will be also put in comparison with existing state-of-the-art synthetic to real domain adaptation methods.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Sampling/Importance Resampling for Semantically Consistent Synthetic to Real Image Domain Adaptation in Urban Traffic Scenes


    Beteiligte:
    Savkin, Artem (Autor:in) / Kasperek, Monika (Autor:in) / Tombari, Federico (Autor:in)


    Erscheinungsdatum :

    2019-06-01


    Format / Umfang :

    3795170 byte





    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Creation of Critical Traffic Scenes for Usage with Importance Sampling

    Jesenski, Stefan / Tiemann, Nils / Branz, Wolfgang et al. | IEEE | 2021



    Real-to-Synthetic: Generating Simulator Friendly Traffic Scenes from Graph Representation

    Tian, Yafu / Carballo, Alexander / Li, Ruifeng et al. | IEEE | 2022


    A Parallel Teacher for Synthetic-to-Real Domain Adaptation of Traffic Object Detection

    Wang, Jiangong / Shen, Tianyu / Tian, Yonglin et al. | IEEE | 2022