...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Journal of circuits, systems and computers >Receptive Field Fusion RetinaNet for Object Detection
【24h】

Receptive Field Fusion RetinaNet for Object Detection

机译：用于物体检测的接收领域融合视网膜

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In modern convolutional neural network (CNN)-based object detector, the extracted features are not suitable for multi-scale detection and all the bounding boxes are simply ranked according to their classification scores in nonmaximum suppression (NMS). To address the above problems, we propose a novel one-stage detector named receptive field fusion RetinaNet. First, receptive field fusion module is proposed to extract richer multi-scale features by fusing feature maps of various receptive fields. Second, joint confidence guided NMS is proposed to optimize the post-processing process of object detection, which introduce location confidence in NMS and take joint confidence as the NMS rank basis. According to our experimental results, significant improvement in terms of mean of average precision (mAP) can be achieved on average compared with the state-of-the-art algorithm.
机译：在现代卷积神经网络（CNN）的对象检测器中，提取的特征不适用于多尺度检测，并且所有边界盒都仅根据其在非抑制（NMS）中的分类分数来排序。 为了解决上述问题，我们提出了一种名为接受现场融合视网膜的新型单级探测器。 首先，提出通过定影各种接收领域的特征图来提取更丰富的多尺度特征的接收领域融合模块。 其次，建议联合信心引导NMS优化物体检测后的后处理过程，这引入了NMS的位置置信度，并以NMS级别为基础而采取联合信心。 根据我们的实验结果，与平均精度（MAP）的平均值的显着改善与最先进的算法相比，可以平均实现。

著录项

  • 来源
    《Journal of circuits, systems and computers》 |2021年第10期|2150184.1-2150184.16|共16页
  • 作者

    Huang He; Feng Yong; Zhou MingLiang; Qiang Baohua; Yan Jielu; Wei Ran;

  • 作者单位

    Chongqing Univ Coll Comp Sci Chongqing 400030 Peoples R China;

    Chongqing Univ Coll Comp Sci Chongqing 400030 Peoples R China;

    Chongqing Univ Key Lab Dependable Serv Comp Cyber Phys Soc Minist Educ Chongqing 400030 Peoples R China;

    Univ Macau State Key Lab Internet Things Smart City Fac Sci & Technol Macau Peoples R China|Guilin Univ Elect Technol Guangxi Key Lab Optoelectron Informat Proc Guilin 541004 Peoples R China|Guilin Univ Elect Technol Guangxi Key Lab Trusted Software Guilin 541004 Peoples R China;

    Univ Macau State Key Lab Internet Things Smart City Fac Sci & Technol Macau Peoples R China;

    Chongqing Med Data Informat Technol Co Ltd Bldg 3 Block B Adm Ctr Chongqing 401336 Peoples R China;

  • 收录信息
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

    Object detection; receptive field; NMS; one-stage detector; multi-scale;

    机译：物体检测;接受场;NMS;一级探测器;多尺度;

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号