|
|
基于候选区域选择及深度网络模型的骑车人识别 |
李晓飞1, 许庆1, 熊辉1,2, 王建强1, 李克强1 |
1. 清华大学 汽车安全与节能国家重点实验室, 北京 100084; 2. 北京航空航天大学 软件学院, 北京 100191 |
|
Cyclist detection based on detection proposals and deep convolutional neural networks |
LI Xiaofei1, XU Qing1, XIONG Hui1,2, WANG Jianqiang1, LI Keqiang1 |
1. State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing 100084, China; 2. School of Software, Beihang University, Beijing 100191, China |
|
摘要基于骑车人目标识别的骑车人保护系统是保护道路环境中骑车人的重要手段。该文提出了骑车人目标的候选区域选择方法,并结合基于深度卷积神经网络的目标分类与定位方法,实现了骑车人目标的有效识别。候选区域选择方法可分为3部分:骑车人共有显著性区域检测、基于冗余策略的候选区域生成和基于车载视觉几何约束的候选区域选择。在公开的骑车人数据库上进行的对比试验表明:相对于现有的目标候选区域选择及目标识别方法,该方法显著提升了骑车人目标的识别率及识别精度,进而验证了该方法的有效性。 |
关键词 :目标识别,骑车人识别,目标候选区域选择,卷积神经网络 |
Abstract:Cyclist protection systems based on cyclist detection methods are needed to protect cyclists from road traffic. This paper presents a detection proposal method and a cyclist detection method using deep convolutional neural networks to classify and locate cyclists. The detection proposal method uses cyclist shared salient region detection, redundancy-based detection and geometric constraint-based detection. Tests using a public cyclist dataset show that this method significantly outperforms state-of-the-art detection proposals, which verifies the effectiveness of this method. |
Key words:object detectioncyclist detectiondetection proposalconvolutional neural network |
收稿日期: 2016-04-24 出版日期: 2017-05-20 |
|
通讯作者:李克强,教授,E-mail:likq@tsinghua.edu.cnE-mail: likq@tsinghua.edu.cn |
[1] | Toroyan T. WHO Global Status Report on Road Safety 2015 [R]. Geneva: World Health Organization, 2015. |
[2] | Geronimo D, Lopez A M, Sappa A D, et al. Survey of pedestrian detection for advanced driver assistance systems [J]. IEEE Trans on PAMI, 2010, 32(7): 1239-1258. |
[3] | LI Tong, CAO Xianbin, XU Yanwu. An effective crossing cyclist detection on a moving vehicle [C]//Proc 8th Intelligent Control and Automation. Jinan, 2010: 368-372. |
[4] | 岳昊, 邵春福, 赵熠. 基于 BP 神经网络的行人和自行车交通识别方法[J]. 北京交通大学学报, 2008, 32(3): 46-49.YUE Hao, SHAO Chunfu, ZHAO Yi. A study on pedestrian and cyclist recognition based on BP neural network [J]. Journal of Beijing Jiaotong University, 2008, 32(3): 46-49. (in Chinese) |
[5] | Cho H, Rybski P E, ZHANG Wende. Vision-based bicyclist detection and tracking for intelligent vehicles [C]//Proc IEEE IV. San Diego, 2010: 454-461. |
[6] | YANG Kai, LIU Chao, ZHENG Jiangyu, et al. Bicyclist detection in large scale naturalistic driving video [C]//Proc IEEE ITSC. Tsingtao, 2014: 1638-1643. |
[7] | TIAN Wei, Lauer M. Fast cyclist detection by cascaded detector and geometric constraint [C]//Proc IEEE ITSC. Canary Islands, 2015: 1286-1291. |
[8] | LI Xiaofei, Flohr F, YANG Yue, et al. A new benchmark for vision-based cyclist detection [C]//Proc IEEE IV. Gothenburg, 2016: 1028-1033. |
[9] | Nam W, Dollár P, Han J H. Local decorrelation for improved pedestrian detection [C]//Proc NIPS. Montreal, 2014: 424-432. |
[10] | Duda R O, Hart P E, Stork D G. Pattern Classification [M]. New York: John Wiley & Sons, 2012. |
[11] | Girshick R, Donahue J, Darrell T, et al. Region-based convolutional networks for accurate object detection and segmentation [J]. IEEE Trans on PAMI, 2016, 38(1): 142-158. |
[12] | Girshick R. Fast R-CNN [C]//Proc IEEE ICCV. Santiago, 2015: 1440-1448. |
[13] | Uijlings J, Sande K, Gevers T, et al. Selective search for object recognition [J]. Int J Comput Vis, 2010, 104(2): 154-171. |
[14] | Dollár P, Zitnick C L. Fast edge detection using structured forests [J]. IEEE Trans on PAMI, 2015, 37(8): 1558-1570. |
[1] | 艾斯卡尔·肉孜, 殷实, 张之勇, 王东, 艾斯卡尔·艾木都拉, 郑方. THUYG-20:免费的维吾尔语语音数据库[J]. 清华大学学报(自然科学版), 2017, 57(2): 182-187. | [2] | 杨莹春, 邓立才. 基于GMM托肯配比相似度校正得分的说话人识别[J]. 清华大学学报(自然科学版), 2017, 57(1): 28-32. | [3] | 聂鼎, 安雪晖. 基于图像处理的净浆扩展度测量工具开发[J]. 清华大学学报(自然科学版), 2016, 56(12): 1249-1254. | [4] | 田垚, 蔡猛, 何亮, 刘加. 基于深度神经网络和Bottleneck特征的说话人识别系统[J]. 清华大学学报(自然科学版), 2016, 56(11): 1143-1148. | [5] | 郑军, 魏海永. 基于白化变换及曲率特征的3维物体识别及姿态计算[J]. 清华大学学报(自然科学版), 2016, 56(10): 1025-1030. | [6] | 陈宝华, 邓磊, 段岳圻, 陈志祥, 周杰. 三维重建中的多模型融合:克服光照和尺度影响[J]. 清华大学学报(自然科学版), 2016, 56(9): 969-973. | [7] | 谢颖, 杨向东, 芮晓飞, 任书楠, 陈恳. 圆柱透视投影轮廓的隐式方程描述和拟合方法[J]. 清华大学学报(自然科学版), 2016, 56(6): 640-645. | [8] | 王晶, 王昊. 融合局部特征和全局特征的视频拷贝检测[J]. 清华大学学报(自然科学版), 2016, 56(3): 269-272. | [9] | 褚洪洋, 柴跃廷, 刘义. 基于层次分裂算法的价格指数序列聚类[J]. 清华大学学报(自然科学版), 2015, 55(11): 1178-1183. | [10] | 路海明, 王一娇, 谢朝霞. 基于向量场的深度计算方法[J]. 清华大学学报(自然科学版), 2015, 55(8): 916-920. | [11] | 杨向东, 芮晓飞, 谢颖. 基于高效Hough变换的圆柱特征检测方法[J]. 清华大学学报(自然科学版), 2015, 55(8): 921-926. | [12] | 姜志威, 丁晓青, 彭良瑞. 针对无切分维吾尔文文本行识别的字符模型优化[J]. 清华大学学报(自然科学版), 2015, 55(8): 873-877,883. | [13] | 肖熙, 王竞千. 基于网格的语音关键词检索算法改进[J]. 清华大学学报(自然科学版), 2015, 55(5): 508-513. |
|