Corpus ID: 210023792

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

@article{Leang2020DynamicTW,
  title={Dynamic Task Weighting Methods for Multi-task Networks in Autonomous Driving Systems},
  author={Isabelle Leang and Ganesh Sistu and Fabian Burger and Andrei Bursuc and Senthil Kumar Yogamani},
  journal={ArXiv},
  year={2020},
  volume={abs/2001.02223}
}
  • Isabelle Leang, Ganesh Sistu, +2 authors Senthil Kumar Yogamani
  • Published in ArXiv 2020
  • Computer Science
  • Deep multi-task networks are of particular interest for autonomous driving systems. They can potentially strike an excellent trade-off between predictive performance, hardware constraints and efficient use of information from multiple types of annotations and modalities. However, training such models is non-trivial and requires balancing the learning of all tasks as their respective losses display different scales, ranges and dynamics across training. Multiple task weighting methods that adjust… CONTINUE READING

    Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 53 REFERENCES

    Multi-task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics

    VIEW 6 EXCERPTS
    HIGHLY INFLUENTIAL

    The Cityscapes Dataset for Semantic Urban Scene Understanding

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL

    Are we ready for autonomous driving? The KITTI vision benchmark suite

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL

    Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation

    VIEW 2 EXCERPTS

    NeurAll: Towards a Unified Visual Perception Model for Automated Driving

    VIEW 1 EXCERPT

    P

    • M. Uřičář
    • Křı́žek, G. Sistu, and S. Yogamani. Soilingnet: Soiling detection on automotive surround-view cameras. In 2019 IEEE Intelligent Transportation Systems Conference (ITSC), pages 67–72. IEEE
    • 2019
    VIEW 1 EXCERPT

    WoodScape: A Multi-Task, Multi-Camera Fisheye Dataset for Autonomous Driving

    VIEW 2 EXCERPTS