International Core Journal of Engineering 2020-26 | Page 155

IV. C ONCLUSION
The detection rate curve is shown in Fig. 9.
In this paper, YOLO algorithm is used to select the target
area of naval vessels in the background of sea and air.
Darknet-53 network is used to record the characteristics of
the target. Combining with the specific application scenarios,
the dimension and objective function of the output tensor of
the model are adjusted, which reduces the model parameters
and improves the efficiency of the model operation. The
experimental results show that the method is effective.
Through the test of 200 images, the ROI extraction algorithm
in this paper can extract the interested objects more
accurately. It has a good extraction effect for all kinds of ship
images with different distances, and the detection rate is
increased by 4.25% on average. This algorithm achieves good
results, but there are still high requirements for hardware. The
next research direction is to simplify the network scale in
order to obtain the optimal cost-effective ratio.
Fig.9
After retraining, the detection rate of YOLO decreased
slightly in the original image, but it increased significantly in
181*181, 128*128 and 86*86 times images. On the original
resolution image, the detection rate decreased by 1%, on
181*181 images, the detection rate increased by 1.5%, on
128*128 images, the detection rate increased by 5%, and on
86*86 images. The output rate increased by 11.5%, indicating
that the more serious the image degradation, the more
obvious the detection rate increased.
R EFERENCES
[1]
[2]
[3]
As shown in Figure 10, the image of the ship position in
the image is labeled by the retrained YOLO.
[4]
[5]
[6]
[7]
Figure 10 Ship Position Detection
133
Chalechale A, Mertins A, Naghdy G. Edge Image Description
Using Angular Radial Partitioning [J]. IEE Proceedings-Vision
Image and Signal Processing, 2004, 151(2): 93-101.
Borba G B, Gamba H R, Marques O. Extraction of Salient Regions of
Interest Using Visual Attention Models [C]. in Proceedings of SPIE
- The International Society for Optical Engineering, San Jose, CA,
United states, 2009, 7255: 247-254.
Wang T, Rui Y, Sun J G. Constraint Based Region Matching for
Image Retrieval [J].International Journal of Computer Vision, 2004,
56(1): 37-45.
Nimsky C, Nelles M, Urbach H. A Standardised Evaluation of Pre-
Surgical Imaging of the Corticospinal Tract: Where to Place the
Seed Roi Comments [J]. Neurosurgical Review, 2009, 32(4): 456-
456.
Lecun Y,Boser B,Denker J S,et al.Backpropagation applied to
handwritten zip code Recognition[J].Neural Computation,1989, 1
(4):541-551.
Redmon J , Divvala S , Girshick R , et al. You Only Look Once: Unified,
Real-Time Object Detection[C]// 2016 IEEE Conference on Computer
Vision and Pattern Recognition (CVPR). IEEE, 2016.
Redmon J , Farhadi A . YOLOv3: An Incremental Improvement[J].
2018.