Unlock instant, AI-driven research and patent intelligence for your innovation.

Information processing device, information processing method, and program

A technology of an information processing device and an information processing method, which is applied in the directions of image data processing, character and pattern recognition, image analysis, etc.

Pending Publication Date: 2022-07-08
CANON KK
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

On the other hand, a model learned by using a learning data set formed of limited data can only cope with recognition target data that changes slightly

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
  • Information processing device, information processing method, and program
  • Information processing device, information processing method, and program
  • Information processing device, information processing method, and program

Examples

Experimental program
Comparison scheme
Effect test

no. 1 example

[0032] It would be advantageous to recognize the recognition target data using a model that has been learned by using a learning dataset with attributes that exactly match the attributes of the recognition target data. If such a learned model exists, it is only necessary to select that model as a suitable model for recognition. However, if the types of attributes are diverse, it is difficult to prepare a model that has been learned by using a learning dataset having attribute patterns corresponding to all combinations. For example, in the case where 10 attributes are set with respect to the data, if the data set is formed by a combination of "include" and "exclude" for each attribute, a data set of 1023 patterns is formed. In this case, if the model is to be learned for each dataset, 1023 learnings must be performed. If the number of attribute types is increased by subdividing data conditions, the number of models to be learned increases further. Since learning usually requi...

no. 2 example

[0094] In the first embodiment, an example of selecting a model suitable for recognizing recognition target data has been explained. However, model selection is not limited to selecting one model, and multiple models can be selected based on diversity. In the second embodiment, an example of selecting a plurality of models suitable for identifying identification target data will be explained.

[0095] Note that the hardware configuration and functional configuration of the information processing apparatus 100 according to the second embodiment are the same as those of the first embodiment, so the explanation thereof will be omitted. Furthermore, as in the first embodiment, an embodiment using a crack identification model for inspecting infrastructure will also be explained in the second embodiment. The processing procedure according to the second embodiment conforms to Figure 4 and Image 6 flow chart shown.

[0096] The model selected in the first embodiment is a model t...

no. 3 example

[0106] In the above-described embodiments, attributes are clear items such as types of structures that humans can visually understand. However, the present invention is not limited to this. In the third embodiment, an example in which attribute information is formed based on image feature amounts will be explained. Note that the hardware configuration and functional configuration of the information processing apparatus 100 according to the third embodiment are the same as those of the first embodiment, so the explanation thereof will be omitted. Furthermore, as in the first embodiment, an embodiment using a crack identification model for inspecting infrastructure will also be explained in the third embodiment.

[0107] In the third embodiment, the information of the attribute set is formed from the image of the learning data set or the image of the recognition target data by a method called Bag-of-Features or Bag-of-Visual Words . Since the feature bag is a well-known metho...

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

An information processing apparatus includes: a first acquisition section configured to acquire information about learning data used in learning of each of a plurality of pre-learning models for identifying input data; a second acquisition unit for acquiring information indicating an attribute for identifying the target data; and a model selection unit configured to select a model based on a degree of matching between an attribute of the recognition target data and an attribute of the learning data used in learning of each of a plurality of models, and on a diversity of the attribute of the learning data used in learning of each of the plurality of models, the recognition target data being recognized based on the degree of matching between the attribute of the recognition target data and the attribute of the learning data used in learning of each of the plurality of models. A model to be used in the recognition of the recognition target data is selected from the plurality of models.

Description

technical field [0001] The present invention relates to an information processing apparatus, an information processing method, and a program, and more particularly, to a technique for selecting a learning model. Background technique [0002] In machine learning, learning models are formed by learning from learning data. If the data content of the learning dataset is different, then a model with different characteristics is formed. When multiple models are learned using multiple different learning datasets as described above, an appropriate model must be selected according to the recognition goal. [0003] PTL 1 discloses a technique by which a model is selected as a model to be used in recognition by using an image having imaging conditions similar to those including the imaging position and imaging angle of the recognition target image to learn. [0004] Furthermore, a model learned by using a learning dataset formed of various kinds of data is generally able to cope wit...

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): G06N20/00G06T7/73
CPCG06V10/87G06V10/945G06T7/0004G06T2207/20081G06T2207/20084G06T2207/30184G06T2207/30132G06F18/285G06F18/2185G06N3/0464G06N3/0475G06N3/045G06N20/20G06V10/7788G06V10/75
Inventor 野上敦史
Owner CANON KK