An Image Annotation Recommendation Method Based on Frequent Itemsets
A technology of frequent item sets and recommendation methods, which is applied in the field of image annotation recommendation based on frequent item sets, can solve problems such as poor efficiency and poor annotation accuracy, and achieve the effect of avoiding multiple calculations
Active Publication Date: 2020-09-01
HOHAI UNIV
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] In order to solve the problems of poor efficiency and poor labeling accuracy in the process of labeling massive pictures in the prior art, the present invention provides a picture labeling recommendation method based on frequent itemsets, which improves the speed and accuracy of picture labeling
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0036] The technical solutions of the present invention will be described in detail below in conjunction with the accompanying drawings.
[0037] The present invention designs a picture annotation recommendation algorithm based on frequent item sets, such as figure 1 shown, including the following steps:
[0038] Step 1: When a new user logs in for the first time, since he has no historical tags, random untagged images will be pushed.
[0039] Step 2: Obtain all labels in the system to form a project set L={l1, l2, l3,...,ln}, where n is the number of all labels. Obtaining the historical tags of each user constitutes a transaction database D={d1,d2,...,dm}, where m is the number of transactions, and each transaction di corresponds to a unique user.
[0040] Step 3: Build a frequent pattern tree. An example of this is as follows:
[0041]In this task, the ID of the transaction is the user, and the item in the transaction is the label record. Substitute specific labels with...
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
Login to View More Abstract
The invention discloses a picture annotation recommendation method based on a frequent item set. The method comprises: firstly, a frequent pattern tree is constructed, and a frequent item set is obtained from the frequent pattern tree; secondly, calculating the coincidence degree, constructing a sparse self-coding neural network, calculating the similarity between pictures in the frequent item setand pictures marked by a user through the sparse self-coding neural network, and then recommending the user; and finally judging whether the picture is marked. According to the invention, frequent item set mining is carried out on the processed and grouped data; a sparse self-coding neural network is added for feature extraction; and according to the extracted features, the picture needing to belabeled is pushed to a user with a corresponding interest or professional knowledge background for labeling, so that the problems of poor efficiency, poor labeling accuracy and the like in a mass picture labeling process are solved, and the picture labeling speed and accuracy are improved.
Description
technical field [0001] The invention belongs to the technical field of data mining, and in particular relates to a method for recommending picture annotations based on frequent item sets. Background technique [0002] During the training process of the machine vision system, it is necessary to gradually improve the accuracy of the machine vision system to identify pictures by identifying the marked samples. In this way, it is necessary to prepare a large number of marked training sample pictures in advance. The traditional method of making training samples is to gather some people and manually label the pictures. This method is not only inefficient, but also because each person has different professional knowledge, resulting in uneven labeling quality. Existing image labeling systems such as LabelImg, BBox-Label-Tool, etc. only support a single user to label pictures, but cannot distribute pictures according to the user's professional knowledge background. Therefore, an inte...
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
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06F16/58G06F16/2458G06F16/9535
Inventor 刘凡吕坦悦
Owner HOHAI UNIV