Weakly-Supervised Domain Adaptation of Deep Regression Trackers via Reinforced Knowledge Distillation

@article{Dunnhofer2021WeaklySupervisedDA,
  title={Weakly-Supervised Domain Adaptation of Deep Regression Trackers via Reinforced Knowledge Distillation},
  author={Matteo Dunnhofer and N. Martinel and C. Micheloni},
  journal={IEEE Robotics and Automation Letters},
  year={2021},
  volume={6},
  pages={5016-5023}
}
Deep regression trackers are among the fastest tracking algorithms available, and therefore suitable for real-time robotic applications. However, their accuracy is inadequate in many domains due to distribution shift and overfitting. In this letter we overcome such limitations by presenting the first methodology for domain adaption of such a class of trackers. To reduce the labeling effort we propose a weakly-supervised adaptation strategy, in which reinforcement learning is used to express… Expand

References

SHOWING 1-10 OF 51 REFERENCES
Visual Tracking by Means of Deep Reinforcement Learning and an Expert Demonstrator
  • 7
  • PDF
SiamRPN++: Evolution of Siamese Visual Tracking With Very Deep Networks
  • 396
  • PDF
Addressing appearance change in outdoor robotics with adversarial domain adaptation
  • 47
  • PDF
Kitting in the Wild through Online Domain Adaptation
  • 24
  • PDF
Adaptive Deep Learning Through Visual Domain Localization
  • 17
  • PDF
Multi-Task Domain Adaptation for Deep Learning of Instance Grasping from Simulation
  • 60
  • PDF
Learning Discriminative Model Prediction for Tracking
  • 167
  • PDF
ATOM: Accurate Tracking by Overlap Maximization
  • 224
  • PDF
...
1
2
3
4
5
...