基于CenterNet的实时行人检测模型  

CenterNet-Based Real-Time Pedestrian Detection Model

作　　者：姜建勇 吴云[1] 龙慧云[1] 黄自萌 蓝林 JIANG Jianyong;WU Yun;LONG Huiyun;HUANG Zimeng;LAN Lin(College of Computer Science and Technology,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学计算机科学与技术学院,贵阳550025

出　　处：《计算机工程》2021年第10期276-282,共7页Computer Engineering

基　　金：国家自然科学基金(61741124);贵州省科技计划项目(5781)。

摘　　要：针对传统目标检测模型不能同时兼顾检测速度和准确度的问题,提出一种新的PD-CenterNet模型。在CenterNet的基础上对网络结构和损失函数进行改进,在网络结构的上采路径中,设计基于注意力机制的特征融合模块,对低级特征和高级特性进行融合,在损失函数中通过设计α、γ、δ3个影响因子来提高正样本与降低负样本的损失,以平衡正负样本的损失。实验结果表明,相比CenterNet模型,该模型在网络结构和损失函数上的准确度分别提高5.1%、9.81%。Generally,the speed gain of traditional target detection models comes at the cost of accuracy,and vice versa.To address the problem,a new pedestrian detection model,PD-CenterNet,is proposed based on CenterNet by improving its network structure and loss function.In terms of network structure,a feature fusion module based on attention mechanism is given in the up-sampling path to fuse low-level features and high-level features.In terms of the loss function,three factorsα、γandδare designed to increase the loss of positive samples and reduce the loss of negative samples,balancing the loss of the samples.Experimental results show that compared with the CenterNet model,the proposed model improves the accuracy of network structure by 5.1%and the accuracy of the loss function by 9.81%.

关 键 词：PD-CenterNet网络 实时检测 行人检测 样本不平衡 损失函数 特征融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]