| 一种基于空洞卷积的遮挡目标检测方法 |
| |
| 引用本文： | 杨少波.一种基于空洞卷积的遮挡目标检测方法[J].软件,2021(1):135-139. |
| |
| 作者姓名： | 杨少波 |
| |
| 作者单位： | ;1.北方工业大学信息学院 |
| |
| 摘 要： | 本文提出了一种基于两阶段目标检测的方法,该方法基于FasterR-CNN模型,以ResNet50为主干网络,利用特征金字塔网络融合多个特征层的上下文信息,并在后续特征图的处理过程中加入空洞卷积,以扩大特征图的感受野,增强对遮挡目标的检测。
|
| 关 键 词： | 卷积神经网络 空洞卷积 遮挡目标检测 |
A Method Base on Dilated Convolution for Occluded Object Detection |
| |
| YANG Shaobo.A Method Base on Dilated Convolution for Occluded Object Detection[J].Software,2021(1):135-139. |
| |
| Authors: | YANG Shaobo |
| |
| Affiliation: | (School of Information Science and Technology,North China University of Technology,Beijing 100144) |
| |
| Abstract: | In this paper,we proposed a two-stage object detection method.The method is based on the Faster R-CNN model,using Res Net50 as the backbone network,and use the feature pyramid networks to reuse the higher-resolution maps of the feature hierarchy.And we add the dilated convolution to abstract the feature map of the last layer of feature maps to expand the receptive field of the feature map and enhance the detection of occluded objects. |
| |
| Keywords: | convolutional neural network dilated convolution obscured object detection |
| 本文献已被 维普 等数据库收录！ |
|
