English

Smooth and Stepwise Self-Distillation for Object Detection

Computer Vision and Pattern Recognition 2024-01-22 v2

Abstract

Distilling the structured information captured in feature maps has contributed to improved results for object detection tasks, but requires careful selection of baseline architectures and substantial pre-training. Self-distillation addresses these limitations and has recently achieved state-of-the-art performance for object detection despite making several simplifying architectural assumptions. Building on this work, we propose Smooth and Stepwise Self-Distillation (SSSD) for object detection. Our SSSD architecture forms an implicit teacher from object labels and a feature pyramid network backbone to distill label-annotated feature maps using Jensen-Shannon distance, which is smoother than distillation losses used in prior work. We additionally add a distillation coefficient that is adaptively configured based on the learning rate. We extensively benchmark SSSD against a baseline and two state-of-the-art object detector architectures on the COCO dataset by varying the coefficients and backbone and detector networks. We demonstrate that SSSD achieves higher average precision in most experimental settings, is robust to a wide range of coefficients, and benefits from our stepwise distillation procedure.

Keywords

Cite

@article{arxiv.2303.05015,
  title  = {Smooth and Stepwise Self-Distillation for Object Detection},
  author = {Jieren Deng and Xin Zhou and Hao Tian and Zhihong Pan and Derek Aguiar},
  journal= {arXiv preprint arXiv:2303.05015},
  year   = {2024}
}

Comments

Accepted by International Conference on Image Processing (ICIP) 2023

R2 v1 2026-06-28T09:08:36.399Z