TY - JOUR
T1 - Object Detection in Aerial Images Using Feature Fusion Deep Networks
AU - Long, Hao
AU - Chung, Yinung
AU - Liu, Zhenbao
AU - Bu, Shuhui
N1 - Funding Information:
This work was supported in part by the Natural Science Foundation of China under Grant 61672430, in part by the Shaanxi Key Research and Development Program under Grant S2019-YF-ZDCXL-ZDLGY-0227, in part by the Aeronautical Science Fund under Grant BK1829-02-3009, in part by the NWPU Basic Research Fund under Grant 3102018jcc001, in part by the Science and Technique Program of Beijing Municipal Education Commission under Grant KM201711417009, and in part by the Ministry of Science and Technology under Grant MOST 105-2221-E-018-023.
PY - 2019
Y1 - 2019
N2 - Object detection acts as an essential part in a wide range of measurement systems in traffic management, urban planning, defense, agriculture, and so on. Convolutional Neural Networks-based researches reach a great improvement on detection tasks in natural scene images enjoying from the strong ability of feature representations. However, because of the high density, the small size of objects, and the intricate background, the current methods achieve relatively low precision in aerial images. The intention of this work is to obtain better detection performance in aerial images by designing a novel deep neural network framework called Feature Fusion Deep Networks (FFDN). The novel architecture combines a designed structural learning layer based on a graphical model. As a result, the network not only provides powerful hierarchical representation but also strengthens the spatial relationship between the high-density objects. We demonstrate the great improvement of the proposed FFDN on the UAV123 data set and another novel challenging data set called UAVDT benchmark. The objects which appear with small size, partial occlusion and out of view, as well as in the dark background can be detected accurately.
AB - Object detection acts as an essential part in a wide range of measurement systems in traffic management, urban planning, defense, agriculture, and so on. Convolutional Neural Networks-based researches reach a great improvement on detection tasks in natural scene images enjoying from the strong ability of feature representations. However, because of the high density, the small size of objects, and the intricate background, the current methods achieve relatively low precision in aerial images. The intention of this work is to obtain better detection performance in aerial images by designing a novel deep neural network framework called Feature Fusion Deep Networks (FFDN). The novel architecture combines a designed structural learning layer based on a graphical model. As a result, the network not only provides powerful hierarchical representation but also strengthens the spatial relationship between the high-density objects. We demonstrate the great improvement of the proposed FFDN on the UAV123 data set and another novel challenging data set called UAVDT benchmark. The objects which appear with small size, partial occlusion and out of view, as well as in the dark background can be detected accurately.
UR - http://www.scopus.com/inward/record.url?scp=85064651668&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85064651668&partnerID=8YFLogxK
U2 - 10.1109/ACCESS.2019.2903422
DO - 10.1109/ACCESS.2019.2903422
M3 - Article
AN - SCOPUS:85064651668
VL - 7
SP - 30980
EP - 30990
JO - IEEE Access
JF - IEEE Access
SN - 2169-3536
M1 - 8661761
ER -