Improved YOLOv3 model side-scan sonar shipwreck target automatic identification method based on transfer learning

A technology of transfer learning and side scan sonar, applied in the field of side scan sonar image target recognition and deep learning, can solve the problems of high false alarm rate, slow recognition speed, poor FasterR-CNN model effect, etc., to reduce leakage The effect of alarm rate, preventing over-fitting and rich detection efficiency

Active Publication Date: 2020-12-08
PLA DALIAN NAVAL ACADEMY
View PDF5 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the above-mentioned deficiencies in the prior art, and propose an improved YOLOv3 model-based side-scan sonar image wreck target recognition method based on transfer learning, so as to solve the existing problems of manual interpretation and manual feature extraction of existing side-scan sonar images , and at the same time solve the problems that the Faster R-CNN model is not effective in small target recognition, has a high false alarm rate and slow recognition speed

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Improved YOLOv3 model side-scan sonar shipwreck target automatic identification method based on transfer learning
  • Improved YOLOv3 model side-scan sonar shipwreck target automatic identification method based on transfer learning
  • Improved YOLOv3 model side-scan sonar shipwreck target automatic identification method based on transfer learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0039] Below the experiment of the present invention is described in detail in conjunction with accompanying drawing:

[0040] The training and testing of this experiment are all based on the TensorFlow framework using python programming. The experimental environment is: Linux: Ubuntu18.04 version operating system; the CPU is Inter(R) Xeon(R) CPU E5-2678 v3@2.50GHz; the GPU is NVIDIA TITAN RTX, 24GB memory.

[0041] This paper is based on the YOLOv3 model, and the specific model structure is as follows figure 2 As shown, the YOLOv3 model uses the Darknet-53 network structure for image feature extraction. As shown in the figure, the network is mainly composed of 53 1×1 and 3×3 convolutional layers (Convolutional), which are located before the Res layer. And each convolutional layer will be followed by a BN layer and a LeakyReLU layer, which together form a DBL, such as figure 2 Shown are the basic components of the YOLOv3 network structure. The YOLOv3 model adds a skip con...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an improved YOLOv3 model side-scan sonar shipwreck target automatic identification method based on transfer learning, and belongs to the technical field of side-scan sonar image target identification and deep learning. The improved YOLOv3 model side-scan sonar image shipwreck target identification method based on transfer learning solves the problems in existing side-scan sonar image manual interpretation and manual feature extraction, and also solves the problems that a Faster R-CNN model is poor in small target recognition effect, high in missing alarm rate and low inrecognition speed. The identification and positioning precision of a shipwreck target is further improved, so that the model achieves a better convergence effect, and finally, the purposes of improving the overall performance of the model and achieving real-time detection are achieved.

Description

technical field [0001] The invention belongs to the technical field of side-scan sonar image target recognition and deep learning, relates to an improved YOLOv3 model-based automatic recognition method for side-scan sonar sunken ship targets based on transfer learning, and is an improved recognition algorithm in the technical field of deep learning target recognition , applied to shipwreck target recognition in side-scan sonar images. Background technique [0002] How to accurately, quickly and efficiently search for wrecked ships is an important part of maritime search and rescue and obstacle verification. Side-scan sonar can be used to detect submarine targets and plays a key role in emergency search and rescue. Side-scan sonar detection generally adopts the towed measurement method. Due to the constraints of sea maneuvering and the length of the towline, the depth of the towed fish is generally only tens of meters. The defect of poor image quality. At present, side-sca...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/20G06V2201/07G06N3/045G06F18/23213G06F18/25
Inventor 金绍华汤寓麟边刚张永厚王美娜
Owner PLA DALIAN NAVAL ACADEMY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap