Commodity positioning detection re-identification method based on feature map

A technology of positioning detection and re-identification, which is applied in the field of image recognition, can solve problems such as undetectable products, and achieve the effects of avoiding product damage that cannot be detected, avoiding product damage, and reducing labor costs

Inactive Publication Date: 2019-09-10
HANGZHOU DIANZI UNIV
View PDF6 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] The present invention adopts the matching strategy of dynamic mask subgraph and parallel feature map, and designs a feature subgraph generation module based on dynamic mask, a convolution mapping and embedding module and a feature re-identification module, which solves the problem that products cannot be detected due to overlapping At the same time, it reduces the impact of other products in the same shopping on the construction of a single product feature subgraph

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
  • Commodity positioning detection re-identification method based on feature map
  • Commodity positioning detection re-identification method based on feature map
  • Commodity positioning detection re-identification method based on feature map

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0065] Such as figure 1 As shown, in this embodiment, a feature map-based commodity location detection and re-identification method includes the following steps:

[0066] 1) Get multi-angle video material sequence {Video k}, according to the "frame-by-frame extraction" method, extract the picture in the video, and get {image k,1 , image k,2 , image k,3 ,...,image k,all}. Use YOLO's ability to quickly locate items to quickly detect image collections {image k} in the position and category of each product, the position of the jth product in the kth frame image is recorded as {j ,Y j > k};

[0067] 2) According to the position information, the position of the jth commodity in the kth frame of the sample image {j ,Y j > k}Generate mask mask matrix M k,j mask , the value of the rectangular box identifying the position of commodity j in the matrix is ​​1, and the rest are 0 (such as figure 2 gray-black area and white area in the middle);

[0068] 3) the M k,j mask Mu...

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 a commodity positioning detection re-identification method based on a feature map. According to the method, a dynamic mask subgraph and parallel feature graph matching method is adopted, the article overlapping detection is optimized, and the commodity recognition efficiency is improved. The method can be used for the commodity identification and settlement work in commodity settlement of supermarkets and sellers, the error rate is reduced, the labor cost is reduced, and the settlement efficiency is improved. By changing the input data, the method can be expanded to beapplied to other article detection fields.

Description

technical field [0001] The invention relates to the field of image recognition, in particular to a feature map-based product location detection and re-recognition method. Background technique [0002] With the development of artificial intelligence, the application of machine vision is becoming more and more extensive. Object classification and detection is a very active research direction in the fields of computer vision, pattern recognition and machine learning. Object classification and detection are widely used in many fields, including face recognition in the security field, pedestrian detection, intelligent video analysis, etc. At the same time, some object detection methods based on CNN and RNN deep neural networks have been born, such as single frame recognition, CNN-based extended network recognition, dual-channel CNN recognition, and LSTM-based recognition methods. However, due to the complex network structure, they still need to be improved in terms of network i...

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): G06K9/00
CPCG06V20/64G06V20/10G06V20/48
Inventor 张剑清俞东进杨宏福孙笑笑
Owner HANGZHOU DIANZI 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