arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.02340cs.CV
收藏

目标检测的通用实例提取

General Instance Distillation for Object Detection

MEGVII Technology(旷视科技);BeiHang University(北京航空航天大学)

Xing Dai, Zeren Jiang, Zhao Wu, Yiping Bao, Zhicheng Wang, Si Liu, Erjin Zhou

摘要:

近年来,知识蒸馏被证明是解决模型压缩问题的有效方法。这种方法可以使轻量级的学生模型从繁琐的教师模型中获取知识。然而,以往的提取检测方法对不同检测框架的泛化能力较弱,严重依赖地面真值(GT),忽略了实例之间有价值的关系信息。因此,我们提出了一种新的蒸馏检测任务的基础上的歧视性的例子,而不考虑的积极或消极区分GT,这就是所谓的一般实例蒸馏(GID)。我们的方法包含一个通用的实例选择模块(GISM),以充分利用基于特征,基于关系和基于响应的知识进行蒸馏。大量的结果表明,学生模型实现了显着的AP改进,甚至在各种检测框架中优于教师。具体来说,RetinaNet与ResNet-50在COCO数据集上使用GID实现了39.1%的mAP,比基线36.2%高出2.9%,甚至优于基于ResNet-101的教师模型的38.1% AP。

英文摘要:

In recent years, knowledge distillation has been proved to be an effective solution for model compression. This approach can make lightweight student models acquire the knowledge extracted from cumbersome teacher models. However, previous distillation methods of detection have weak generalization for different detection frameworks and rely heavily on ground truth (GT), ignoring the valuable relation information between instances. Thus, we propose a novel distillation method for detection tasks based on discriminative instances without considering the positive or negative distinguished by GT, which is called general instance distillation (GID). Our approach contains a general instance selection module (GISM) to make full use of feature-based, relation-based and response-based knowledge for distillation. Extensive results demonstrate that the student model achieves significant AP improvement and even outperforms the teacher in various detection frameworks. Specifically, RetinaNet with ResNet-50 achieves 39.1% in mAP with GID on COCO dataset, which surpasses the baseline 36.2% by 2.9%, and even better than the ResNet-101 based teacher model with 38.1% AP.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑