Residual Super-Resolution Single Shot Network for Low-Resolution Object Detection

被引:25
作者
Zhao, Xiaotong [1 ,2 ]
Li, Wei [3 ]
Zhang, Yifan [1 ,2 ]
Feng, Zhiyong [1 ]
机构
[1] Beijing Univ Posts & Telecommun, Key Lab Universal Wireless Commun, Beijing 100876, Peoples R China
[2] Beijing Univ Posts & Telecommun, Res Inst, Shenzhen 518057, Peoples R China
[3] Northern Illinois Univ, Dept Elect Engn, De Kalb, IL 60115 USA
来源
IEEE ACCESS | 2018年 / 6卷
基金
中国国家自然科学基金;
关键词
Object detection; convolutional neural networks; image resolution; IMAGE SUPERRESOLUTION; RECOGNITION;
D O I
10.1109/ACCESS.2018.2867586
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
For object detection in computer vision, detection models trained by high-resolution images often fail to recognize or localize objects on low-resolution images. To tackle this problem, we propose a fully convolutional network named residual super-resolution single shot network (RSRSSN). RSRSSN consists of two sub-networks, super-resolution sub-network and detection sub-network. The super-resolution sub-network in RSRSSN is achieved by stacking of identity residual blocks while the detection sub-network adopts the single shot multibox detector (SSD). Based on multi-task learning, we design a novel objective function called feature maps multibox loss to enforce low-resolution images to produce similar feature maps with their corresponding high-resolution ones. This information sharing mechanism is proved to be critical for solving the resolution mismatch problem in the experiments. A two-step training scheme is also proposed to train the RSRSSN in an end-to-end manner. Without any data augmentation, RSRSSN outperforms the SSD on both down-sampled PASCAL VOC and MS COCO in real-time object detection.
引用
收藏
页码:47780 / 47793
页数:14
相关论文
共 51 条
[1]  
[Anonymous], PROC CVPR IEEE
[2]  
[Anonymous], 2014, VERY DEEP CONVOLUTIO
[3]  
[Anonymous], 2017, IEEE I CONF COMP VIS, DOI DOI 10.1109/ICCV.2017.322
[4]  
[Anonymous], PATTERN RECOGNIT LET
[5]  
[Anonymous], IEEE T PATTERN ANAL
[6]  
[Anonymous], PROC CVPR IEEE
[7]  
[Anonymous], 2010, P 27 INT C INT C MAC
[8]  
[Anonymous], IEEE T PATTERN ANAL
[9]  
[Anonymous], P INT C NEUR INF PRO
[10]   Saliency-Based Pedestrian Detection in Far Infrared Images [J].
Cai, Yingfeng ;
Liu, Ze ;
Wang, Hai ;
Sun, Xiaoqiang .
IEEE ACCESS, 2017, 5 :5013-5019