TY - GEN
T1 - TransNet
T2 - 2019 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2019
AU - Zhang, Weitong
AU - Tian, Yanling
AU - Zhang, Qieshi
AU - Cheng, Jun
AU - Zheng, Huiquan
N1 - Funding Information:
∗Corresponding author: qs.zhang@siat.ac.cn This work was supported by National Key R&D Program of China (2018YFB1308000), National Natural Science Funds of China (U1813205, U1713213, 61772508, 61801428), Guangdong Technology Project (2017B010110007, 2016B010108010), Zhejiang Provincial Natural Science Foundation of China (LY18F020034, LY18F020032) Shenzhen Technology Project (JCYJ20180507182610734, JCYJ20170413152535587), CAS Key Technology Talent Program, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences (2014DP173025), Shenzhen Engineering Laboratory for 3D Content Generating Technologies (NO. [2017]476).
Funding Information:
This work was supported by National Key R&D Program of China (2018YFB1308000), National Natural Science Funds of China (U1813205, U1713213, 61772508, 61801428), Guangdong Technology Project (2017B010110007, 2016B010108010), Zhejiang Provincial Natural Science Foundation of China (LY18F020034, LY18F020032) Shenzhen Technology Project (JCYJ20180507182610734, JCYJ20170413152535587), CAS Key Technology Talent Program, Key Laboratory of Human-Machine Intelligence-Synergy Systems, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences (2014DP173025), Shenzhen Engineering Laboratory for 3D Content Generating Technologies (NO. [2017]476).
Publisher Copyright:
© 2019 IEEE.
PY - 2019/8
Y1 - 2019/8
N2 - One of the major challenges in object recognition is to propose a structure with highly performance of overlapping, small and misclassification objects, called hard object recognition. There is a contextual nexus among the spatial location and geometric characteristics of objects, which we find is essential to the recognition efficacy. We propose regress connection, nexus calculation module and linkage loss, which are contained in an improving design TransNet framework. It can not only demonstrates the interactivity of explicit nexus between objects, but also achieve better results through the integration of two mainstream methods. Adjusting the training attention through linkage loss function, thus significantly enhance the performance of hard objects. The experiment results show that our method achieves remarkable performance on PASCAL VOC2012 and MS COCO data set. Also, the head network is capable of integrating with region-based methods and gains better performance.
AB - One of the major challenges in object recognition is to propose a structure with highly performance of overlapping, small and misclassification objects, called hard object recognition. There is a contextual nexus among the spatial location and geometric characteristics of objects, which we find is essential to the recognition efficacy. We propose regress connection, nexus calculation module and linkage loss, which are contained in an improving design TransNet framework. It can not only demonstrates the interactivity of explicit nexus between objects, but also achieve better results through the integration of two mainstream methods. Adjusting the training attention through linkage loss function, thus significantly enhance the performance of hard objects. The experiment results show that our method achieves remarkable performance on PASCAL VOC2012 and MS COCO data set. Also, the head network is capable of integrating with region-based methods and gains better performance.
UR - http://www.scopus.com/inward/record.url?scp=85083316356&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85083316356&partnerID=8YFLogxK
U2 - 10.1109/RCAR47638.2019.9044144
DO - 10.1109/RCAR47638.2019.9044144
M3 - Conference contribution
AN - SCOPUS:85083316356
T3 - 2019 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2019
SP - 75
EP - 79
BT - 2019 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2019
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 4 August 2019 through 9 August 2019
ER -