This article introduces a versatile multi-task learning framework (UMT-Net) and an adaptive task weighting (ATW) training method, specifically designed for resource-constrained scenarios that demand parameter-efficient networks. The adaptable UMT-Net architecture includes a global-shared backbone based on an encoder, task-specific self-attention modules, inter-task joint-attention fusion modules, and feature-aggregating decoders. The ATW technique accounts for both short-term variation and long-term statistics of task losses, leading to a more stable training process. Extensive experiments on CityScapes and NYUv2 datasets reveal that UMT-Net outperforms baseline methods while requiring fewer computations, model parameters, and inference latency. In addition, we also conducted experiments on the autonomous driving dataset BDD100 K and achieved state-of-the-art performance. Furthermore, we deployed the model and carried out tests for generalization in real-world scenarios. Finally, our network architecture possesses the capability to be designed as a compact model with much fewer parameters, computational requirements, and inference time, while maintaining competitive performance, making it suitable for deployment on mobile devices.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    UMT-Net: A Uniform Multi-Task Network With Adaptive Task Weighting


    Contributors:
    Chen, Sihan (author) / Zheng, Lianqing (author) / Huang, Libo (author) / Bai, Jie (author) / Zhu, Xichan (author) / Ma, Zhixiong (author)

    Published in:

    Publication date :

    2024-01-01


    Size :

    25075697 byte




    Type of media :

    Article (Journal)


    Type of material :

    Electronic Resource


    Language :

    English




    Dynamic Task Weighting Methods for Multi-task Networks in Autonomous Driving Systems

    Leang, Isabelle / Sistu, Ganesh / Burger, Fabian et al. | IEEE | 2020


    Multiple metric learning with query adaptive weights and multi-task re-weighting for person re-identification

    Jia, Jieru / Ruan, Qiuqi / An, Gaoyun et al. | British Library Online Contents | 2017


    MtpNet: Multi-Task Panoptic Driving Perception Network

    Li, Zheng / Yuan, Xiaohui / Sun, Bifan et al. | IEEE | 2025


    An Adaptive Task Planning Method for UAVC Task Layer: DSTCA

    Ting Duan / Qun Li / Xin Zhou et al. | DOAJ | 2024

    Free access