We address the image retrieval problem for a wireless system including an edge server and an edge device. The query image is first compressed by the edge device, and then transmitted into wireless channel, while the edge server retrieves the received image. Different from conventional schemes directly compressing features via unsupervised learning regardless of the database semantic distribution, we design a deep semantic coding (DSC) scheme by integrating the inverted semantic index structure of the database into the coding process, which can utilize the prior semantic information of the database to reduce the bandwidth. We extract the feature vectors from the images via a convolutional neural network and generate the semantic guided code head, which is followed by the product quantization. The experimental results verify the effectiveness of the DSC scheme in reducing the bandwidth as well as improving the performance of wireless image retrieval.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Deep Semantic Coding for Wireless Image Retrieval


    Beteiligte:
    Wang, Ying (Autor:in) / Qi, Chenhao (Autor:in)


    Erscheinungsdatum :

    2022-09-01


    Format / Umfang :

    788794 byte





    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch




    Semantic kernel learning for interactive image retrieval

    Gosselin, P.H. / Cord, M. | IEEE | 2005


    Semantic Kernel Learning for Interactive Image Retrieval

    Gosselin, P. H. / Cord, M. | British Library Conference Proceedings | 2005


    Image Retrieval with Well-Separated Semantic Hash Centers

    Wang, Liangdao / Pan, Yan / Lai, Hanjiang et al. | British Library Conference Proceedings | 2023


    Hidden semantic concept discovery in region based image retrieval

    Ruofei Zhang, / Zhongfei Zhang, | IEEE | 2004