[1]马浩良,谢林柏.基于 SSD 的不平衡样本车辆检测与识别[J].计算机技术与发展,2019,29(12):135-140.[doi:10. 3969 / j. issn. 1673-629X. 2019. 12. 024]
　MA Hao-liang,XIE Lin-bo.Vehicle Detection and Recognition in Unbalanced Samples Based on SSD[J].,2019,29(12):135-140.[doi:10. 3969 / j. issn. 1673-629X. 2019. 12. 024]
点击复制

基于 SSD 的不平衡样本车辆检测与识别()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
29
期数:
2019年12期
页码:
135-140
栏目:
应用开发研究
出版日期:
2019-12-10

文章信息/Info

Title:
Vehicle Detection and Recognition in Unbalanced Samples Based on SSD
文章编号:
1673-629X(2019)12-0135-06
作者:
马浩良; 谢林柏
江南大学 物联网工程学院,江苏 无锡 214122
Author(s):
MA Hao-liang; XIE Lin-bo
School of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China
关键词:
车辆检测与识别; SSD; 样本不平衡; 难易样本挖掘; 正负样本挖掘
Keywords:
vehicle detection and recognition; SSD; sample imbalance; hard sample mining; positive and negative sample mining
分类号:
TP301
DOI:
10. 3969 / j. issn. 1673-629X. 2019. 12. 024
摘要:
为了实现在复杂环境,车辆样本不平衡情况下的实时车辆检测与识别,基于 SSD 算法搭建了车辆检测与识别的框架。 针对车辆数据存在车型难易样本不均衡以及 SSD 方法存在的正负样本不平衡问题,在 SSD 引入改进的损失函数来挖掘难易样本,通过提高难样本的学习比例来更好地识别样本较少的车辆类型。 引入 SSD 级联的网络结构,在第一级 SSD 挖掘正负样本,在第二级 SSD 根据第一级 SSD 的指导过滤掉大量的负样本。 构建了拥有 7 480 幅图像,包含4 种车辆类型的数据集对该方法进行验证。 实验结果表明,基于改进 SSD 的方法提高了少样本车辆类型的准确率,使整体检测精度取得了 90.0%的准确率。 针对不均衡样本的车辆数据集有较好的通用性,适用于车辆检测与识别任务。
Abstract:
In order to realize real-time vehicle detection and recognition in a complex environment with unbalanced vehicle samples,a framework of vehicle detection and recognition is built based on SSD algorithm. Aiming at the imbalance of difficult and easy samples in vehicle data and the imbalance of positive and negative samples in SSD method,an improved loss function to mine difficult and easy samples in SSD is introduced to identify vehicle types with fewer samples better by increasing the learning proportion of difficult samples. The cascade SSD network structure is introduced to mine positive and negative samples in the first-level SSD and filter out a large number of negative samples in the second-level SSD according to the guidance of the first-level SSD. A data set with 7 480 images and 4 vehicle types is onstructed to verify this method. The experiment shows that the improved method based on SSD improves the accuracy of vehicle types with fewer samples and the overall detection accuracy achieves 90.0%. This method has excellent generality for vehicle data sets with unbalanced samples and is suitable for vehicle detection and recognition.

相似文献/References:

[1]艾丽蓉,李凯.基于Postgres-XL 的数据管理优化技术研究[J].计算机技术与发展,2018,28(03):11.[doi:TP311]
　AI Li-rong,LI Kai.Study on Optimization Technology of Data Management Based on Postgres-XL[J].,2018,28(12):11.[doi:TP311]
[2]杜华,刘华春.云数据中心下重复数据删除技术研究[J].计算机技术与发展,2019,29(02):157.[doi:10．3969/j．issn．1673－629X．2019．02．033]
　DU Hua,LIU Huachun.Research on Deduplication of Data in Cloud Data Center[J].,2019,29(12):157.[doi:10．3969/j．issn．1673－629X．2019．02．033]
[3]唐小敏,舒远仲,刘文祥,等.基于 SSD 深度网络的河道漂浮物检测技术研究[J].计算机技术与发展,2020,30(09):154.[doi:10. 3969 / j. issn. 1673-629X. 2020. 09. 028]
　TANG Xiao-min,SHU Yuan-zhong,LIU Wen-xiang,et al.Research on River Floating Object Detection Technology Based on SSD Deep Network[J].,2020,30(12):154.[doi:10. 3969 / j. issn. 1673-629X. 2020. 09. 028]
[4]邓 权,林明星.基于改进 SSD 的海洋生物检测算法[J].计算机技术与发展,2022,32(04):51.[doi:10. 3969 / j. issn. 1673-629X. 2022. 04. 009]
　DENG Quan,LIN Ming-xing.Marine Organism Detection Algorithm Based on Improved SSD[J].,2022,32(12):51.[doi:10. 3969 / j. issn. 1673-629X. 2022. 04. 009]

更新日期/Last Update: 2019-12-10