Image salient target detection method combined with deep learning

A deep learning and target detection technology, applied in the field of image processing, can solve the problems of lack of prominence of high-level semantic features, redundant features, and too much discrete noise.

Inactive Publication Date: 2018-11-27
SOUTHWEST JIAOTONG UNIV
View PDF5 Cites 49 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Compared with traditional methods, learning-based methods have further breakthroughs in performance, but there are also problems s

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
  • Image salient target detection method combined with deep learning
  • Image salient target detection method combined with deep learning
  • Image salient target detection method combined with deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0045] Example: Image Salient Object Detection

[0046] parameter settings

[0047] In this paper, based on the improved neural network model of VGG16, the MSRA-B data set, which is widely used in salient target detection, is selected. [17] As a training set, it contains 2500 natural scene images and their corresponding artificially labeled truth maps. The scene semantics are diverse, and it comes from the MSRA5000 data set. Input the original image and its corresponding true value image into the network model for training, each initial parameter is set as: basic learning rate 10 -8 , the weight attenuation coefficient is 0.0005, the momentum is 0.9, the number of batches is set to 1, the initial maximum number of iterations is set to 15,000, and the entire neural network is trained using the "SGD" learning rate attenuation method. The initial parameters set by the neural network model are used for training and iterative optimization to solve θ. When the number of iterations...

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 provides an image salient target detection method combined with deep learning. The method is based on an improved RFCN deep convolution neural network of cross-level feature fusion, anda network model comprises two parts of basic feature extraction and cross-level feature fusion. The method comprises: firstly, using an improved deep convolution network model to extract features of an input image, and using a cross-level fusion framework for feature fusion, to generate a high-level semantic feature preliminary saliency map; then, fusing the preliminary saliency map with image bottom-layer features to perform saliency propagation and obtain structure information; finally, using a conditional random field (CRF) to optimize a saliency propagation result to obtain a final saliency map. In a PR curve graph obtain by the method, F value and MAE effect are better than those obtained by other nine algorithms. The method can improve integrity of salient target detection, and has characteristics of less background noise and high algorithm robustness.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image detection method based on a deep learning computer mathematical model. Background technique [0002] In today's era of Internet popularization and explosion of image information, it is becoming more and more difficult for human beings to process external image information only through their own visual perception system, and using computers to process image information has become an effective method. The researchers simulated the visual perception mechanism that humans can quickly pay attention to the region of interest in the image, and proposed an image salient object detection method. Due to the ability to extract the key information of the image scene, the amount of calculation is greatly reduced when performing subsequent processing under limited resources. [1] , image scene understanding [2] , target perception [3] , image and video compression [4] and ...

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
IPC IPC(8): G06K9/46G06K9/62G06K9/34
CPCG06V10/267G06V10/462G06F18/23G06F18/253G06F18/214
Inventor 安维胜赵恒
Owner SOUTHWEST JIAOTONG UNIV
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