Assisted Excitation of Activations: A Learning Technique to Improve Object Detectors

@article{Derakhshani2019AssistedEO,
  title={Assisted Excitation of Activations: A Learning Technique to Improve Object Detectors},
  author={Mohammad Mahdi Derakhshani and S. Masoudnia and Amir Hossein Shaker and Omid Mersa and M. Sadeghi and M. Rastegari and B. Araabi},
  journal={2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2019},
  pages={9193-9202}
}
  • Mohammad Mahdi Derakhshani, S. Masoudnia, +4 authors B. Araabi
  • Published 2019
  • Computer Science
  • 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • We present a simple yet effective learning technique that significantly improves mAP of YOLO object detectors without compromising their speed. During network training, we carefully feed in localization information. We excite certain activations in order to help the network learn to better localize (Figure 2). In the later stages of training, we gradually reduce our assisted excitation to zero. We reached a new state-of-the-art in the speed-accuracy trade-off (Figure 1). Our technique improves… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 36 REFERENCES
    YOLOv3: An Incremental Improvement
    2949
    You Only Look Once: Unified, Real-Time Object Detection
    8287
    Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
    11072
    Going deeper with convolutions
    19284
    Feature Pyramid Networks for Object Detection
    3920
    Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
    17591
    YOLO9000: Better, Faster, Stronger
    4195
    Focal Loss for Dense Object Detection
    2148
    Single-Shot Object Detection with Enriched Semantics
    97